feat: init 'machine rm' command that outputs service containers that will be removed

* test: Add an e2e test for build&push

* test: Add image cleanup

* test: Switch to assert.NoError

.dockerignore

@@ -4,6 +4,7 @@
 # Allow files and directories.
 !cmd/
 !internal/
+!pkg/
 !scripts/docker/
 !go.*
 

.github/FUNDING.yml

@@ -0,0 +1 @@
+github: [psviderski]

.github/workflows/go-tests.yml

@@ -0,0 +1,67 @@
+name: Go Tests
+on:
+  push:
+    branches:
+      - "main"
+      - "test/**"
+      - "release/**"
+  pull_request:
+    branches:
+      - main
+    paths:
+      - ".github/**"
+      - "**.go"
+      - "**.proto"
+      - "go.*"
+      - "Makefile"
+      - "scripts/**"
+      - "test/**"
+permissions:
+  contents: read
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+
+      - name: Set up Go
+        uses: actions/setup-go@d35c59abb061a4a6fb18e82ac0862c26744d6ab5 # v5.5.0
+        with:
+          go-version: "1.23.2"
+
+      - name: Install dependencies
+        run: |
+          go mod tidy
+          git diff --exit-code ||
+            (echo "go.mod or go.sum has changed. Please run 'go mod tidy' and commit the changes." && exit 1)
+
+      - name: Run tests
+        run: make test
+    timeout-minutes: 10
+
+  check-protobuf:
+    runs-on: ${{ matrix.os }}
+    timeout-minutes: 10
+    strategy:
+      matrix:
+        # Run on more platforms so that mise lockfile checksums are checked/generated for all platforms.
+        # See https://github.com/jdx/mise/issues/4276 for details.
+        os: [ubuntu-latest, macos-latest]
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+
+      - name: Install Mise
+        uses: jdx/mise-action@13abe502c30c1559a5c37dff303831bab82c9402 # v2.2.3
+        with:
+          version: "2025.6.5"
+        env:
+          GITHUB_TOKEN: ${{ github.token }}
+
+      - name: Generate Protobuf files
+        run: |
+          make proto
+          # check if the generated files are up to date
+          git diff --exit-code ||
+            (echo "Generated protobuf files are not up to date. Please run 'make proto' and commit the changes." && exit 1)

.gitignore

@@ -27,8 +27,10 @@ go.work.sum
 .env
 .idea/
 
-data
+/data/
-dist/
+/dist/
+/uc
+/tmp
 
 # Web
 node_modules/

.goreleaser.yaml

@@ -24,6 +24,8 @@ builds:
     goarch:
       - amd64
       - arm64
+    ldflags:
+      - -s -w -X github.com/psviderski/uncloud/internal/version.version={{ .Version }}
 
   - id: uncloudd
     main: ./cmd/uncloudd
@@ -35,6 +37,8 @@ builds:
     goarch:
       - amd64
       - arm64
+    ldflags:
+      - -s -w -X github.com/psviderski/uncloud/internal/version.version={{ .Version }}
 
 archives:
   - id: uncloud

.mise.lock

@@ -0,0 +1,31 @@
+[tools.go]
+version = "1.23.10"
+backend = "core:go"
+
+[tools.go.checksums]
+"go1.23.10.darwin-arm64.tar.gz" = "sha256:25c64bfa8a8fd8e7f62fb54afa4354af8409a4bb2358c2699a1003b733e6fce5"
+"go1.23.10.linux-amd64.tar.gz" = "sha256:535f9f81802499f2a7dbfa70abb8fda3793725fcc29460f719815f6e10b5fd60"
+
+[tools.protoc]
+version = "27.3"
+backend = "aqua:protocolbuffers/protobuf/protoc"
+
+[tools.protoc.checksums]
+"protoc-27.3-linux-x86_64.zip" = "sha256:6dab2adab83f915126cab53540d48957c40e9e9023969c3e84d44bfb936c7741"
+"protoc-27.3-osx-aarch_64.zip" = "sha256:b22116bd97cdbd7ea25346abe635a9df268515fe5ef5afa93cd9a68fc2513f84"
+
+[tools.protoc-gen-go]
+version = "1.34.2"
+backend = "aqua:protocolbuffers/protobuf-go/protoc-gen-go"
+
+[tools.protoc-gen-go.checksums]
+"protoc-gen-go.v1.34.2.darwin.arm64.tar.gz" = "sha256:17aca7f948dbb624049030cf841e35895cf34183ba006e721247fdeb95ff2780"
+"protoc-gen-go.v1.34.2.linux.amd64.tar.gz" = "sha256:b87bc134dee55576a842141bf0ed27761c635d746780fce5dee038c6dd16554f"
+
+[tools.protoc-gen-go-grpc]
+version = "1.5.1"
+backend = "aqua:grpc/grpc-go/protoc-gen-go-grpc"
+
+[tools.protoc-gen-go-grpc.checksums]
+"protoc-gen-go-grpc.v1.5.1.darwin.arm64.tar.gz" = "sha256:d6083feb51dcfe59f26793e99ef01ef5eac68b64294ec2546711f614ac5878f3"
+"protoc-gen-go-grpc.v1.5.1.linux.amd64.tar.gz" = "sha256:a6cac4ea731e54aea304ad44d704a69d1cdc82997084b33637e21a89dc9229d6"

.mise.toml

@@ -1,5 +1,11 @@
+[settings]
+experimental = true
+
 [tools]
 go = "1.23"
+protoc = "27.3"
+protoc-gen-go = "1.34.2"
+protoc-gen-go-grpc = "1.5.1"
 
 [env]
 _.file = ".env"

Dockerfile

@@ -17,13 +17,22 @@ RUN GOOS=${TARGETOS} GOARCH=${TARGETARCH} go build -o uncloudd cmd/uncloudd/main
 
 
 FROM alpine:${ALPINE_VERSION} AS corrosion-download
-RUN wget -q -O /tmp/corrosion.tar.gz \
+ARG TARGETARCH
-      https://github.com/psviderski/corrosion/releases/latest/download/corrosion-aarch64-unknown-linux-gnu.tar.gz \
+
+RUN CORROSION_ARCH=$(case "${TARGETARCH}" in \
+      "amd64") echo "x86_64" ;; \
+      "arm64") echo "aarch64" ;; \
+      *) echo "Architecture '${TARGETARCH}' not supported" >&2 && exit 1 ;; \
+    esac) \
+    && wget -q -O /tmp/corrosion.tar.gz \
+      "https://github.com/psviderski/corrosion/releases/latest/download/corrosion-${CORROSION_ARCH}-unknown-linux-gnu.tar.gz" \
     && tar -xzf /tmp/corrosion.tar.gz -C /tmp \
     && install /tmp/corrosion /usr/local/bin/corrosion \
     && rm /tmp/corrosion.tar.gz /tmp/corrosion
 
-FROM chainguard/wolfi-base:latest AS corrosion
+# Beware that more modern images like chainguard/wolfi-base build glibc with flags that require newer CPU features,
+# e.g. x86-64-v2. So such image may fail with "Fatal glibc error: CPU does not support x86-64-v2" on older CPUs.
+FROM gcr.io/distroless/cc-debian12:latest AS corrosion
 COPY --from=corrosion-download /usr/local/bin/corrosion /usr/local/bin/corrosion
 CMD ["corrosion", "agent"]
 

Makefile

@@ -1,5 +1,6 @@
 CORROSION_IMAGE ?= ghcr.io/psviderski/corrosion:latest
 UCIND_IMAGE ?= ghcr.io/psviderski/ucind:latest
+DOCS_IMAGE ?= ghcr.io/psviderski/uncloud-docs:latest
 
 update-dev:
 	GOOS=linux GOARCH=amd64 go build -o uncloudd-linux-amd64 ./cmd/uncloudd && \
@@ -30,6 +31,16 @@ reset-dev:
 	ssh spy@192.168.40.176 "sudo systemctl stop uncloud && sudo rm -rf /var/lib/uncloud"
 	ssh ubuntu@152.67.101.197 "sudo systemctl stop uncloud && sudo rm -rf /var/lib/uncloud"
 
+demo-reset:
+	rm -fv ~/.config/uncloud/config.yaml
+	ssh ubuntu@152.67.101.197 "AUTO_CONFIRM=true sudo -E uncloud-uninstall && docker rmi caddy:2.9.1"
+	ssh root@5.223.45.199 "AUTO_CONFIRM=true sudo -E uncloud-uninstall"
+	ssh spy@192.168.40.243 "AUTO_CONFIRM=true sudo -E uncloud-uninstall"
+
+.PHONY: ucind-cluster
+ucind-cluster:
+	go run ./cmd/ucind cluster rm && go run ./cmd/ucind cluster create -m 3
+
 .PHONY: proto
 proto:
 	protoc --go_out=. --go_opt=paths=source_relative --go-grpc_out=. --go-grpc_opt=paths=source_relative \
@@ -39,14 +50,45 @@ proto:
 corrosion-image:
 	docker build -t "$(CORROSION_IMAGE)" --target corrosion .
 
-.PHONY: corrosion-image-push
+.PHONY: corrosion-multiarch-image-push
-corrosion-image-push: corrosion-image
+corrosion-multiarch-image-push:
-	docker push "$(CORROSION_IMAGE)"
+	docker buildx build --push --platform linux/amd64,linux/arm64 -t "$(CORROSION_IMAGE)" --target corrosion .
 
 .PHONY: ucind-image
 ucind-image:
 	docker build -t "$(UCIND_IMAGE)" --target ucind .
 
-.PHONY: ucind-image-push
+.PHONY: ucind-multiarch-image-push
-ucind-image-push: ucind-image
+ucind-multiarch-image-push:
-	docker push "$(UCIND_IMAGE)"
+	docker buildx build --push --platform linux/amd64,linux/arm64 -t "$(UCIND_IMAGE)" --target ucind .
+
+.PHONY: test
+test:
+ifeq ($(TEST_NAME),)
+	go test -count=1 -v ./...
+else
+	go test -count=1 -v -run ^$(TEST_NAME)$$ ./...
+endif
+
+.PHONY: test-clean
+test-clean:
+	@CONTAINERS=$$(docker ps --filter "name=ucind-test" -q); \
+	if [ -n "$$CONTAINERS" ]; then \
+		echo "Killing containers..."; \
+		docker kill $$CONTAINERS; \
+	fi; \
+	CONTAINERS_STOPPED=$$(docker ps -a --filter "name=ucind-test" -q); \
+	if [ -n "$$CONTAINERS_STOPPED" ]; then \
+		echo "Removing stopped containers..."; \
+		docker rm $$CONTAINERS_STOPPED; \
+	else \
+		echo "Nothing to clean."; \
+	fi
+
+.PHONY: vet
+vet:
+	go vet ./...
+
+.PHONY: docs-image-push
+docs-image:
+	docker buildx build --push --platform linux/amd64,linux/arm64 -t "$(DOCS_IMAGE)" ./docs

README.md

@@ -2,47 +2,70 @@
   <img src="./website/images/logo.svg" height="100" alt="Uncloud logo"/>
   <h1>Uncloud</h1>
   <p><strong>Docker simplicity. Multi-machine power.</strong></p>
+
+  <p>
+    <a href="https://docs.uncloud.run"><img src="https://img.shields.io/badge/Docs-blue.svg?style=for-the-badge&logo=gitbook&logoColor=white" alt="Documentation"></a>
+    <a href="https://discord.gg/eR35KQJhPu"><img src="https://img.shields.io/badge/discord-5865F2.svg?style=for-the-badge&logo=discord&logoColor=white" alt="Join Discord"></a>
+    <a href="https://x.com/psviderski"><img src="https://img.shields.io/badge/follow-black?style=for-the-badge&logo=X&logoColor=while" alt="Follow on X"></a>
+    <a href="https://github.com/sponsors/psviderski"><img src="https://img.shields.io/badge/Donate-EA4AAA.svg?style=for-the-badge&logo=githubsponsors&logoColor=white" alt="Donate"></a>
+  </p>
 </div>
 
 Uncloud is a lightweight clustering and container orchestration tool that lets you deploy and manage web apps across
-cloud VMs and bare metal with minimized cluster management overhead. It creates a secure WireGuard mesh network between
+cloud VMs and bare metal with minimised cluster management overhead. It creates a secure WireGuard mesh network between
 your Docker hosts and provides automatic service discovery, load balancing, ingress with HTTPS, and simple CLI commands
 to manage your apps.
 
 Unlike traditional orchestrators, there's no central control plane and quorum to maintain. Each machine maintains a
-synchronized copy of the cluster state through peer-to-peer communication, keeping cluster operations functional even if
+synchronised copy of the cluster state through peer-to-peer communication, keeping cluster operations functional even if
 some machines go offline.
 
-Uncloud aims to be the solution for developers who want the flexibility of self-hosted infrastructure without the
+Uncloud is the solution for developers who want the flexibility of self-hosted infrastructure without the operational
-operational complexity of Kubernetes.
+complexity of Kubernetes.
 
 ## ✨ Features
 
 * **Deploy anywhere**: Combine cloud VMs, dedicated servers, and bare metal into a unified computing environment —
   regardless of location or provider.
+* **Docker Compose**: Familiar [Docker Compose](https://compose-spec.io/) format for defining services and volumes. No
+  need to learn a new bespoke DSL.
+* **Zero-downtime deployments**: Rolling updates without service interruption. Automatic rollback on failure is coming
+  soon.
+* **Service discovery**: Built-in DNS server resolves service names to container IPs.
+* **Persistent storage**: Run stateful services with Docker volumes managed across machines.
 * **Zero-config private network**: Automatic WireGuard mesh with peer discovery and NAT traversal. Containers get unique
   IPs for direct cross-machine communication.
-* **No control plane**: Fully decentralized design eliminates single points of failure and reduces operational overhead.
+* **No control plane**: Fully decentralised design eliminates single points of failure and reduces operational overhead.
 * **Imperative over declarative**: Favoring imperative operations over state reconciliation simplifies both the mental
   model and troubleshooting.
-* **Service discovery**: Built-in DNS server resolves service names to container IPs.
+* **Managed DNS**: Automatic DNS records `*.<id>.cluster.uncloud.run` for services with public access via managed
+  [Uncloud DNS](https://github.com/psviderski/uncloud-dns) service.
 * **Automatic HTTPS**: Built-in Caddy reverse proxy handles TLS certificate provisioning and renewal using Let's
   Encrypt.
 * **Docker-like CLI**: Familiar commands for managing both infrastructure and applications.
 * **Remote management**: Control your entire infrastructure through SSH access to any single machine in the cluster.
 
-Coming soon:
+### 🚀 Coming soon
 
-* Project isolation through environments/namespaces
+* **[Unregistry](https://github.com/psviderski/unregistry) integration**: Push your Docker images directly to your
-* Infrastructure as Code using Docker Compose format
+  machines without an external registry. It will transfer only the missing layers, making it fast and efficient.
-* Persistent volumes and secrets management
-* Monitoring and log aggregation
-* Database deployment and management
-* Curated application catalog
 
-Here is a diagram of an Uncloud multi-provider cluster of 3 machines:
+## 🎬 Quick demo
 
-![Diagram: multi-provider cluster of 3 machines](website/images/diagram.webp)
+The screenshot below demonstrates how I use Uncloud to deploy the [Uncloud Documentation](https://docs.uncloud.run)
+website to 2 remote machines (why not?) from the [`compose.yaml`](docs/compose.yaml) file on my local machine.
+
+It exposes the container port `8000/tcp` as HTTPS on the domain `docs.uncloud.run`, served by the Caddy reverse proxy on
+the remote machines. All managed by Uncloud.
+
+![Uncloud compose deployment demo](.github/images/compose-deploy.jpg)
+
+Here is a more advanced use case. Deploy a highly available web app with automatic HTTPS across multiple regions and
+on-premises in just a couple minutes.
+
+<a href="https://uncloud.wistia.com/medias/k47uwt9uau?wvideo=k47uwt9uau">
+<img src="https://embed-ssl.wistia.com/deliveries/3cf7014a48b93afc556444bed3e39a8c.jpg?image_crop_resized=900x526&image_play_button_rounded=true&image_play_button_size=2x&image_play_button_color=18181Be0" alt="Uncloud demo" width="450" height="263" />
+</a>
 
 ## 💫 Why Uncloud?
 
@@ -52,60 +75,111 @@ believe there's a sweet spot in between — a pragmatic solution for the majorit
 scale. You should be able to:
 
 * **Own your infrastructure and data**: Whether driven by costs, compliance, or flexibility, run applications on any
-  combination of cloud VMs and personal hardware while maintaining the cloud-like experience you love.
+  combination of cloud VMs and personal hardware while controlling your data and maintaining the cloud-like experience
-* **Stay simple**: Don't worry about control planes, highly-available clusters, or complex YAML configurations for
+  you love.
-  common use cases.
+* **Stay simple as you grow**: Start with a single machine and add more whenever you need without changing your
+  workflow. No worrying about highly-available control planes or complex YAML configurations.
 * **Build with proven primitives**: Get production-grade networking, deployment primitives, service discovery, load
   balancing, and ingress with HTTPS out of the box without becoming a distributed systems expert.
-* **Support sustainable computing** 🌿: Minimize system overhead to maximize resources available for your applications.
+* **Support sustainable computing** 🌿: Minimise system overhead to maximise resources available for your applications.
 
-Uncloud's goal is to make deployment and management of containerized applications feel as seamless as using a cloud
+Uncloud's goal is to make deployment and management of containerised applications feel as seamless as using a cloud
 platform, whether you're running on a $5 VPS, a spare Mac mini, or a rack of bare metal servers.
 
 ## 🚀 Quick start
 
 1. Install Uncloud CLI:
 
-```bash
+   ```bash
-brew install psviderski/tap/uncloud
+   brew install psviderski/tap/uncloud
 
-# or using curl (macOS/Linux)
+   # or using curl (macOS/Linux)
-curl -fsS https://get.uncloud.run/install.sh | sh
+   curl -fsS https://get.uncloud.run/install.sh | sh
-```
+   ```
 
-2. Initialize your first machine:
+2. Initialise your first machine:
 
-```bash
+   ```bash
-uc machine init root@your-server-ip
+   uc machine init root@your-server-ip
-```
+   ```
 
-3. Create a DNS A record in your DNS provider (Cloudflare, Namecheap, etc.) that points `app.example.com` to your
+3. Deploy your app from a Docker image and publish its container port 8000 as HTTPS using `app.example.com` domain:
+
+   ```bash
+   uc run -p app.example.com:8000/https image/my-app
+   ```
+
+4. Create a DNS A record in your DNS provider (Cloudflare, Namecheap, etc.) that points `app.example.com` to your
    server's IP address. Allow a few minutes for DNS propagation.
-4. Deploy your app from a Docker image:
 
-```bash
+   That's it! Your app is now running and accessible at https://app.example.com ✨
-uc run -p app.example.com:8000/https my-app-image
-```
-
-That's it! Your app is now running and accessible at https://app.example.com ✨
 
 5. Clean up when you're done:
 
-```bash
+   ```bash
-uc ls
+   uc ls
-# Copy the service ID from the output and remove it:
+   # Copy the service name from the output and run the rm command:
-uc rm my-app-id
+   uc rm my-app-name
-```
+   ```
+
+   If you want to fully uninstall Uncloud on a machine, run:
+
+   ```bash
+   uncloud-uninstall
+   ```
+
+View the [Documentation](https://docs.uncloud.run) for more information.
 
 ## ⚙️ How it works
 
-Check out the [design document](docs/design.md) to understand Uncloud's design philosophy and goals. Here, let's peek
+Check out the [design document](docs/design.md) to understand Uncloud's design philosophy and goals.
-under the hood to see what happens when you run certain commands.
 
-**When you initialize a new cluster on a machine:**
+Here is a diagram of an Uncloud multi-provider cluster of 3 machines:
+
+![Diagram: multi-provider cluster of 3 machines](website/images/diagram.webp)
+
+<details>
+<summary>Peek under the hood to see what happens when you run certain commands.</summary>
+
+**When you initialise a new cluster on a machine:**
 
 ```bash
-uc machine init root@your-server-ip
+$ uc machine init --name oracle-vm ubuntu@152.67.101.197
+
+Downloading Uncloud install script: https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/install.sh
+⏳ Running Uncloud install script...
+✓ Docker is already installed.
+⏳ Installing Docker...
+...
+✓ Docker installed successfully.
+✓ Linux user and group 'uncloud' created.
+✓ Linux user 'ubuntu' added to group 'uncloud'.
+⏳ Installing Uncloud binaries...
+⏳ Downloading uncloudd binary: https://github.com/psviderski/uncloud/releases/latest/download/uncloudd_linux_arm64.tar.gz
+✓ uncloudd binary installed: /usr/local/bin/uncloudd
+⏳ Downloading uninstall script: https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/uninstall.sh
+✓ uncloud-uninstall script installed: /usr/local/bin/uncloud-uninstall
+✓ Systemd unit file created: /etc/systemd/system/uncloud.service
+Created symlink /etc/systemd/system/multi-user.target.wants/uncloud.service → /etc/systemd/system/uncloud.service.
+⏳ Downloading uncloud-corrosion binary: https://github.com/psviderski/corrosion/releases/latest/download/corrosion-aarch64-unknown-linux-gnu.tar.gz
+✓ uncloud-corrosion binary installed: /usr/local/bin/uncloud-corrosion
+✓ Systemd unit file created: /etc/systemd/system/uncloud-corrosion.service
+⏳ Starting Uncloud machine daemon (uncloud.service)...
+✓ Uncloud machine daemon started.
+✓ Uncloud installed on the machine successfully! 🎉
+Cluster "default" initialised with machine "oracle-vm"
+Waiting for the machine to be ready...
+
+Reserved cluster domain: xuw3xd.cluster.uncloud.run
+[+] Deploying service caddy 1/1
+ ✔ Container caddy-c47x on oracle-vm  Started                                                                                                                                          0.9s
+
+Updating cluster domain records in Uncloud DNS to point to machines running caddy service...
+[+] Verifying internet access to caddy service 1/1
+ ✔ Machine oracle-vm (152.67.101.197)  Reachable                                                                                                                                       0.1s
+
+DNS records updated to use only the internet-reachable machines running caddy service:
+  *.xuw3xd.cluster.uncloud.run  A → 152.67.101.197
 ```
 
 1. The CLI SSHs into the machine and installs Docker, the `uncloudd` machine daemon and
@@ -121,7 +195,42 @@ uc machine init root@your-server-ip
 **When you add another machine:**
 
 ```bash
-uc machine add ubuntu@second-server-ip
+$ uc machine add --name hetzner-server root@5.223.45.199
+Downloading Uncloud install script: https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/install.sh
+⏳ Running Uncloud install script...
+✓ Docker is already installed.
+✓ Linux user and group 'uncloud' created.
+⏳ Installing Uncloud binaries...
+⏳ Downloading uncloudd binary: https://github.com/psviderski/uncloud/releases/latest/download/uncloudd_linux_amd64.tar.gz
+✓ uncloudd binary installed: /usr/local/bin/uncloudd
+⏳ Downloading uninstall script: https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/uninstall.sh
+✓ uncloud-uninstall script installed: /usr/local/bin/uncloud-uninstall
+✓ Systemd unit file created: /etc/systemd/system/uncloud.service
+Created symlink /etc/systemd/system/multi-user.target.wants/uncloud.service → /etc/systemd/system/uncloud.service.
+⏳ Downloading uncloud-corrosion binary: https://github.com/psviderski/corrosion/releases/latest/download/corrosion-x86_64-unknown-linux-gnu.tar.gz
+✓ uncloud-corrosion binary installed: /usr/local/bin/uncloud-corrosion
+✓ Systemd unit file created: /etc/systemd/system/uncloud-corrosion.service
+⏳ Starting Uncloud machine daemon (uncloud.service)...
+✓ Uncloud machine daemon started.
+✓ Uncloud installed on the machine successfully! 🎉
+Machine "hetzner-server" added to cluster
+Waiting for the machine to be ready...
+
+[+] Deploying service caddy 1/1
+ ✔ Container caddy-d36c on hetzner-server  Started                                                                                                                                     1.0s
+
+Updating cluster domain records in Uncloud DNS to point to machines running caddy service...
+[+] Verifying internet access to caddy service 2/2
+ ✔ Machine hetzner-server (5.223.45.199)  Reachable                                                                                                                                    0.2s
+ ✔ Machine oracle-vm (152.67.101.197)     Reachable                                                                                                                                    0.1s
+
+DNS records updated to use only the internet-reachable machines running caddy service:
+  *.xuw3xd.cluster.uncloud.run  A → 152.67.101.197, 5.223.45.199
+
+$ uc machine ls
+NAME             STATE   ADDRESS         PUBLIC IP        WIREGUARD ENDPOINTS
+oracle-vm        Up      10.210.0.1/24   152.67.101.197   10.0.0.95:51820, 152.67.101.197:51820
+hetzner-server   Up      10.210.1.1/24   5.223.45.199     5.223.45.199:51820, [2a01:4ff:2f0:128b::1]:51820
 ```
 
 1. The second machine gets provisioned just like the first. A non-root SSH user will need `sudo` access.
@@ -140,7 +249,14 @@ and automatically establish a WireGuard tunnel with it.
 **When you run a service:**
 
 ```bash
-uc run -p app.example.com:8000/https my-app-image
+$ uc run -p app.example.com:8000/https image/my-app
+
+[+] Running service my-app-1b3b (replicated mode) 1/1
+ ✔ Container my-app-1b3b-tcex on oracle-vm  Started
+
+my-app-1b3b endpoints:
+ • https://app.example.com → :8000
+ • https://my-app-1b3b.xuw3xd.cluster.uncloud.run → :8000
 ```
 
 1. CLI picks a machine to run your container.
@@ -158,23 +274,52 @@ Look ma, no control plane or master nodes to maintain! Just a simple overlay net
 sync that lets machines work together. Want to check on things or make changes? Connect to any machine either implicitly
 using the CLI or directly over SSH. They all have the complete cluster state and can control everything. It's like each
 machine is a full backup of your control plane.
+</details>
 
 ## 🏗 Project status
 
 Uncloud is currently in active development and is **not ready for production use**. Features may change significantly
 and there may be breaking changes between releases.
 
-I'd love your input! Here's how you can contribute:
+We'd love your input! Here's how you can contribute:
 
 * 🐛 Found a bug? [Open an issue](https://github.com/psviderski/uncloud/issues)
-* 💡 Have ideas? [Join the discussion](https://github.com/psviderski/uncloud/discussions)
+* 💡 Have questions, ideas, or need help?
+    * Start a discussion or join an existing one in
+      the [Discussions](https://github.com/psviderski/uncloud/discussions).
+    * Join our [Discord community](https://discord.gg/eR35KQJhPu) where we discuss features, roadmap, implementation
+      details, and help each other out.
+
+## 🙏 Inspiration & Acknowledgements
+
+I'm grateful to the following projects that inspired Uncloud's design and implementation:
+
+* [Kamal](https://kamal-deploy.org/) — for proving that even in the declarative era of Kubernetes there is a place for
+  simple deployment tools that use imperative commands without complex orchestration. Kamal powers the multi-billion
+  dollar company [37signals](https://37signals.com/) where it was created, and that's truly inspiring!
+* [Fly.io](https://fly.io/) — for inspiring my vision for what self-hosted infrastructure should feel like, proving that
+  developer experience and powerful infrastructure can coexist beautifully.
+* [Tailscale](https://tailscale.com/) — for pioneering the vision of decentralised flat mesh networking with an amazing
+  user experience that feels like magic.
+* [Talos Linux](https://github.com/siderolabs/talos)
+  and [KubeSpan](https://www.talos.dev/v1.10/talos-guides/network/kubespan/) — for the machine API design using
+  [grpc-proxy](https://github.com/siderolabs/grpc-proxy) and for its elegant approach to secure WireGuard-based overlay
+  networking with zero configuration.
+* [Docker Swarm Classic](https://github.com/docker-archive/classicswarm) and
+  [Rancher 1.x](http://rancher-com-website-main-elb-elb-1798790864.us-west-2.elb.amazonaws.com/docs/rancher/v1.6/en/)
+  — for showing the power of simplicity and pragmatism in container orchestration and that not every problem needs the
+  complexity of Kubernetes.
+
+Special thanks to the [Corrosion](https://github.com/superfly/corrosion) project by Fly.io for providing the distributed
+SQLite database used to share Uncloud's cluster state.
 
 ## 📫 Stay updated
 
-* [Subscribe](https://uncloud.run/#subscribe) to my newsletter to follow the progress, get early insights into new
+* Join our [Discord server](https://discord.gg/eR35KQJhPu) for real-time discussions, support, and updates.
+* Follow [@psviderski](https://x.com/psviderski) on X/Twitter.
+* Subscribe to [my newsletter](https://uncloud.run/#subscribe) to follow the progress, get early insights into new
   features, and be the first to know when it's ready for production use.
 * Watch this repository for releases.
-* Follow [@psviderski](https://github.com/psviderski) on GitHub.
 
 ## ❤️ Contributors
 

cmd/ucind/cluster/create.go

@@ -3,7 +3,7 @@ package cluster
 import (
 	"fmt"
 	"github.com/spf13/cobra"
-	"uncloud/internal/ucind"
+	"github.com/psviderski/uncloud/internal/ucind"
 )
 
 func NewCreateCommand() *cobra.Command {

cmd/ucind/cluster/remove.go

@@ -3,7 +3,7 @@ package cluster
 import (
 	"fmt"
 	"github.com/spf13/cobra"
-	"uncloud/internal/ucind"
+	"github.com/psviderski/uncloud/internal/ucind"
 )
 
 func NewRemoveCommand() *cobra.Command {

cmd/ucind/main.go

@@ -3,12 +3,13 @@ package main
 import (
 	"context"
 	"fmt"
-	"github.com/docker/docker/client"
-	"github.com/spf13/cobra"
 	"os"
 	"strings"
-	"uncloud/cmd/ucind/cluster"
+
-	"uncloud/internal/ucind"
+	"github.com/docker/docker/client"
+	"github.com/psviderski/uncloud/cmd/ucind/cluster"
+	"github.com/psviderski/uncloud/internal/ucind"
+	"github.com/spf13/cobra"
 )
 
 func main() {
@@ -44,9 +45,9 @@ func main() {
 	}
 
 	// TODO: allow to override using UNCLOUD_CONFIG env var.
-	cmd.PersistentFlags().StringVar(&configPath, "uncloud-config", "~/.config/uncloud/config.toml",
+	cmd.PersistentFlags().StringVar(&configPath, "uncloud-config", "~/.config/uncloud/config.yaml",
 		"path to the Uncloud configuration file.")
-	_ = cmd.MarkPersistentFlagFilename("uncloud-config", "toml")
+	_ = cmd.MarkPersistentFlagFilename("uncloud-config", "yaml", "yml")
 
 	cmd.AddCommand(
 		cluster.NewRootCommand(),

cmd/uncloud/build.go

@@ -0,0 +1,77 @@
+package main
+
+import (
+	"context"
+	"fmt"
+
+	composecli "github.com/compose-spec/compose-go/v2/cli"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/client/compose"
+	"github.com/spf13/cobra"
+)
+
+// NewBuildCommand creates a new command to build services from a Compose file.
+func NewBuildCommand() *cobra.Command {
+	opts := cli.BuildOptions{}
+	cmd := &cobra.Command{
+		Use:   "build [FLAGS] [SERVICE...]",
+		Short: "Build services from a Compose file.",
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+
+			if len(args) > 0 {
+				opts.Services = args
+			}
+
+			return runBuild(cmd.Context(), uncli, opts)
+		},
+	}
+
+	cmd.Flags().StringSliceVarP(&opts.Files, "file", "f", nil,
+		"One or more Compose files to build (default compose.yaml)")
+	cmd.Flags().StringSliceVarP(&opts.Profiles, "profile", "p", nil,
+		"One or more Compose profiles to enable.")
+	cmd.Flags().BoolVarP(&opts.Push, "push", "P", false,
+		"Push built images to the registry after building. (default false)")
+	cmd.Flags().BoolVarP(&opts.NoCache, "no-cache", "n", false,
+		"Do not use cache when building images. (default false)")
+
+	return cmd
+}
+
+// TODO: deduplicate with a similar functino for deploy options
+// projectOpts returns the project options for the Compose file(s).
+func projectOptsFromBuildOpts(opts cli.BuildOptions) []composecli.ProjectOptionsFn {
+	projectOpts := []composecli.ProjectOptionsFn{}
+
+	if len(opts.Profiles) > 0 {
+		projectOpts = append(projectOpts, composecli.WithDefaultProfiles(opts.Profiles...))
+	}
+
+	return projectOpts
+}
+
+// runBuild parses the Compose file(s), builds the services, and pushes them if requested.
+func runBuild(ctx context.Context, uncli *cli.CLI, opts cli.BuildOptions) error {
+	projectOpts := projectOptsFromBuildOpts(opts)
+	project, err := compose.LoadProject(ctx, opts.Files, projectOpts...)
+	if err != nil {
+		return fmt.Errorf("load compose file(s): %w", err)
+	}
+
+	if len(opts.Services) > 0 {
+		project, err = project.WithSelectedServices(opts.Services)
+		if err != nil {
+			return fmt.Errorf("select services: %w", err)
+		}
+	}
+
+	servicesToBuild := cli.GetServicesThatNeedBuild(project)
+
+	if len(servicesToBuild) == 0 {
+		fmt.Println("No services to build.")
+		return nil
+	}
+
+	return cli.BuildServices(ctx, servicesToBuild, opts)
+}

cmd/uncloud/caddy/deploy.go

@@ -4,22 +4,23 @@ import (
 	"context"
 	"errors"
 	"fmt"
-	"github.com/charmbracelet/huh"
-	"github.com/docker/cli/cli/streams"
-	"github.com/docker/compose/v2/pkg/progress"
-	"github.com/spf13/cobra"
 	"maps"
 	"slices"
 	"strings"
-	"uncloud/internal/cli"
+
-	"uncloud/internal/cli/client"
+	"github.com/docker/cli/cli/streams"
-	"uncloud/internal/machine/api/pb"
+	"github.com/docker/compose/v2/pkg/progress"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/psviderski/uncloud/pkg/client"
+	"github.com/spf13/cobra"
 )
 
 type deployOptions struct {
-	image   string
+	image    string
-	machine string
+	machines []string
-	cluster string
+	context  string
 }
 
 func NewDeployCommand() *cobra.Command {
@@ -32,24 +33,25 @@ func NewDeployCommand() *cobra.Command {
 			"A rolling update is performed when updating existing containers to minimise disruption.",
 		RunE: func(cmd *cobra.Command, args []string) error {
 			uncli := cmd.Context().Value("cli").(*cli.CLI)
-			return deploy(cmd.Context(), uncli, opts)
+			return runDeploy(cmd.Context(), uncli, opts)
 		},
 	}
 
 	cmd.Flags().StringVar(&opts.image, "image", "",
 		"Caddy Docker image to deploy. (default caddy:LATEST_VERSION)")
-	cmd.Flags().StringVarP(&opts.machine, "machine", "m", "",
+	cmd.Flags().StringSliceVarP(&opts.machines, "machine", "m", nil,
-		"Machine names to deploy to (comma-separated). (default is all machines)")
+		"Machine names to deploy to. Can be specified multiple times or as a comma-separated "+
+			"list of machine names. (default is all machines)")
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.context, "context", "c", "",
-		"Name of the cluster to deploy to. (default is the current cluster)",
+		"Name of the cluster context to deploy to. (default is the current context)",
 	)
 
 	return cmd
 }
 
-func deploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
+func runDeploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
-	clusterClient, err := uncli.ConnectCluster(ctx, opts.cluster)
+	clusterClient, err := uncli.ConnectCluster(ctx, opts.context)
 	if err != nil {
 		return fmt.Errorf("connect to cluster: %w", err)
 	}
@@ -57,7 +59,7 @@ func deploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
 
 	svc, err := clusterClient.InspectService(ctx, client.CaddyServiceName)
 	if err != nil {
-		if !errors.Is(err, client.ErrNotFound) {
+		if !errors.Is(err, api.ErrNotFound) {
 			return fmt.Errorf("inspect caddy service: %w", err)
 		}
 		fmt.Printf("Service: %s (not running)\n", client.CaddyServiceName)
@@ -86,74 +88,46 @@ func deploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
 	fmt.Println()
 	fmt.Println("Preparing a deployment plan...")
 
-	var filter client.MachineFilter
+	placement := api.Placement{
-	if opts.machine != "" {
+		Machines: cli.ExpandCommaSeparatedValues(opts.machines),
-		machines := strings.Split(opts.machine, ",")
-		for i, m := range machines {
-			machines[i] = strings.TrimSpace(m)
-		}
-		filter = machineFilter(machines)
 	}
-
+	d, err := clusterClient.NewCaddyDeployment(opts.image, placement)
-	d, err := clusterClient.NewCaddyDeployment(opts.image, filter)
 	if err != nil {
 		return fmt.Errorf("create caddy deployment: %w", err)
 	}
 
-	// Initialize a machine and container name resolver to properly format the plan output.
-	machines, err := clusterClient.ListMachines(ctx)
-	if err != nil {
-		return fmt.Errorf("list machines: %w", err)
-	}
-	machineNames := make(map[string]string, len(machines))
-	for _, m := range machines {
-		machineNames[m.Machine.Id] = m.Machine.Name
-	}
-	containerNames := make(map[string]string, len(svc.Containers))
-	for _, c := range svc.Containers {
-		containerNames[c.Container.ID] = c.Container.NameWithoutSlash()
-	}
-	resolver := client.NewNameResolver(machineNames, containerNames)
-
 	if opts.image == "" {
 		fmt.Printf("Target image: %s (latest stable)\n", d.Spec.Container.Image)
 	}
 
 	plan, err := d.Plan(ctx)
 	if err != nil {
-		if errors.Is(err, client.ErrNoMatchingMachines) {
-			return fmt.Errorf("no machines found matching: %s", opts.machine)
-		}
 		return fmt.Errorf("plan caddy deployment: %w", err)
 	}
 
-	if len(plan.SequenceOperation.Operations) == 0 {
+	if len(plan.Operations) == 0 {
-		if opts.machine != "" {
+		fmt.Printf("%s service is up to date.\n", client.CaddyServiceName)
-			fmt.Printf("%s service is up to date on selected machines.\n", client.CaddyServiceName)
-		} else {
-			fmt.Printf("%s service is up to date.\n", client.CaddyServiceName)
-		}
 	} else {
 		if svc.ID == "" {
-			if opts.machine != "" {
+			if len(opts.machines) > 0 {
-				fmt.Println("This will run a Caddy container on selected machines.")
+				fmt.Println("This will run a Caddy container on each selected machine.")
 			} else {
 				fmt.Println("This will run a Caddy container on each machine.")
 			}
-		} else {
-			if opts.machine != "" {
-				fmt.Println("This will perform a rolling update of Caddy containers on selected machines.")
-			} else {
-				fmt.Println("This will perform a rolling update of Caddy containers on each machine.")
-			}
 		}
-		fmt.Println()
 
+		// Initialise a machine and container name resolver to properly format the plan output.
+		resolver, err := clusterClient.ServiceOperationNameResolver(ctx, svc)
+		if err != nil {
+			return fmt.Errorf("create machine and container name resolver for service operations: %w", err)
+		}
+
+		fmt.Println()
 		fmt.Println("Deployment plan:")
 		fmt.Println(plan.Format(resolver))
 		fmt.Println()
 
-		confirmed, err := confirm()
+		confirmed, err := cli.Confirm()
 		if err != nil {
 			return fmt.Errorf("confirm deployment: %w", err)
 		}
@@ -178,8 +152,9 @@ func deploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
 }
 
 func UpdateDomainRecords(ctx context.Context, clusterClient *client.Client, progressOut *streams.Out) error {
-	if _, err := clusterClient.GetDomain(ctx); err != nil {
+	domain, err := clusterClient.GetDomain(ctx)
-		if errors.Is(err, client.ErrNotFound) {
+	if err != nil {
+		if errors.Is(err, api.ErrNotFound) {
 			fmt.Println("Skipping DNS records update as no cluster domain is reserved (see 'uc dns').")
 			return nil
 		}
@@ -190,7 +165,7 @@ func UpdateDomainRecords(ctx context.Context, clusterClient *client.Client, prog
 	// TODO: split the method into two: one to get the records and one to update them to ask for update confirmation.
 
 	var records []*pb.DNSRecord
-	err := progress.RunWithTitle(ctx, func(ctx context.Context) error {
+	err = progress.RunWithTitle(ctx, func(ctx context.Context) error {
 		var err error
 		records, err = clusterClient.CreateIngressRecords(ctx, client.CaddyServiceName)
 		return err
@@ -198,18 +173,19 @@ func UpdateDomainRecords(ctx context.Context, clusterClient *client.Client, prog
 	if err != nil {
 		if errors.Is(err, client.ErrNoReachableMachines) {
 			fmt.Println()
-			fmt.Println("DNS records could not be updated as there are no internet-reachable machines running " +
+			fmt.Printf("DNS records for domain '%s' could not be updated as there are no internet-reachable "+
-				"caddy containers.")
+				"machines running caddy containers.\n", domain)
 			fmt.Println()
 			fmt.Println("Possible solutions:")
 			fmt.Println("- Ensure your machines have public IP addresses")
 			fmt.Println("- Use --public-ip flag when adding machines to override the automatically detected IPs")
 			fmt.Println("- Check firewall settings on your machines")
 			fmt.Println("- Configure port forwarding if behind NAT")
-			fmt.Println("- Retry Caddy deployment after resolving connectivity issues with 'uc caddy deploy'")
+			fmt.Println("- Retry Caddy deployment with 'uc caddy deploy' after resolving connectivity issues")
 			fmt.Println()
-			fmt.Println("Your services will not be accessible from the internet until at least one machine " +
+			fmt.Println("Your services won't be accessible from the internet until at least one machine " +
-				"becomes reachable.")
+				"becomes reachable. If you aren't planning to expose any services publicly, you can release " +
+				"the domain by running 'uc dns release'.")
 		}
 		return fmt.Errorf("failed to update DNS records pointing to caddy service: %w", err)
 	}
@@ -222,32 +198,3 @@ func UpdateDomainRecords(ctx context.Context, clusterClient *client.Client, prog
 
 	return nil
 }
-
-func confirm() (bool, error) {
-	var confirmed bool
-	form := huh.NewForm(
-		huh.NewGroup(
-			huh.NewConfirm().
-				Title(
-					"Do you want to continue?",
-				).
-				Affirmative("Yes!").
-				Negative("No").
-				Value(&confirmed),
-		),
-	)
-	if err := form.Run(); err != nil {
-		return false, err
-	}
-
-	return confirmed, nil
-}
-
-func machineFilter(machines []string) client.MachineFilter {
-	if len(machines) == 0 {
-		return nil
-	}
-	return func(m *pb.MachineInfo) bool {
-		return slices.Contains(machines, m.Name)
-	}
-}

cmd/uncloud/context/ls.go

@@ -0,0 +1,61 @@
+package context
+
+import (
+	"fmt"
+	"maps"
+	"os"
+	"slices"
+	"text/tabwriter"
+
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/spf13/cobra"
+)
+
+func NewListCommand() *cobra.Command {
+	var clusterContext string
+
+	cmd := &cobra.Command{
+		Use:     "ls",
+		Aliases: []string{"list"},
+		Short:   "List available cluster contexts.",
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+			return list(uncli)
+		},
+	}
+
+	cmd.Flags().StringVarP(
+		&clusterContext, "context", "c", "",
+		"Name of the cluster context. (default is the current context)",
+	)
+
+	return cmd
+}
+
+func list(uncli *cli.CLI) error {
+	if uncli.Config == nil {
+		return fmt.Errorf("context management is not available: Uncloud configuration file is not being used")
+	}
+
+	if len(uncli.Config.Contexts) == 0 {
+		fmt.Println("No contexts found")
+		return nil
+	}
+
+	contextNames := slices.Sorted(maps.Keys(uncli.Config.Contexts))
+	currentContext := uncli.Config.CurrentContext
+
+	tw := tabwriter.NewWriter(os.Stdout, 0, 0, 3, ' ', 0)
+	fmt.Fprintln(tw, "NAME\tCURRENT\tCONNECTIONS")
+
+	for _, name := range contextNames {
+		current := ""
+		if name == currentContext {
+			current = "*"
+		}
+		connCount := len(uncli.Config.Contexts[name].Connections)
+		fmt.Fprintf(tw, "%s\t%s\t%d\n", name, current, connCount)
+	}
+
+	return tw.Flush()
+}

cmd/uncloud/context/root.go

@@ -0,0 +1,25 @@
+package context
+
+import (
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/spf13/cobra"
+)
+
+func NewRootCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:     "ctx",
+		Aliases: []string{"context"},
+		Short:   "Switch between different cluster contexts. Contains subcommands to manage contexts.",
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+			return selectContext(uncli)
+		},
+	}
+
+	cmd.AddCommand(
+		NewListCommand(),
+		NewUseCommand(),
+	)
+
+	return cmd
+}

cmd/uncloud/context/use.go

@@ -0,0 +1,82 @@
+package context
+
+import (
+	"fmt"
+	"maps"
+	"slices"
+
+	"github.com/charmbracelet/huh"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/spf13/cobra"
+)
+
+func NewUseCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "use [CONTEXT]",
+		Short: "Switch to a different cluster context.",
+		Long: "Switch to a different cluster context. If no context is provided, " +
+			"a list of available contexts will be displayed for selection.",
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+
+			if len(args) == 1 {
+				if err := uncli.SetCurrentContext(args[0]); err != nil {
+					return fmt.Errorf("failed to set the current cluster context to '%s': %w", args[0], err)
+				}
+				fmt.Printf("Current cluster context is now '%s'.\n", args[0])
+				return nil
+			}
+
+			return selectContext(uncli)
+		},
+	}
+
+	return cmd
+}
+
+func selectContext(uncli *cli.CLI) error {
+	if uncli.Config == nil {
+		return fmt.Errorf("context management is not available: Uncloud configuration file is not being used")
+	}
+	if len(uncli.Config.Contexts) == 0 {
+		return fmt.Errorf("no contexts found in Uncloud config (%s)", uncli.Config.Path())
+	}
+
+	contextNames := slices.Sorted(maps.Keys(uncli.Config.Contexts))
+
+	var selected string
+	form := huh.NewForm(
+		huh.NewGroup(
+			huh.NewSelect[string]().
+				Title("Select a cluster context").
+				Options(buildContextOptions(contextNames, uncli.Config.CurrentContext)...).
+				Value(&selected),
+		),
+	)
+	if err := form.Run(); err != nil {
+		return fmt.Errorf("select cluster context: %w", err)
+	}
+
+	if err := uncli.SetCurrentContext(selected); err != nil {
+		return fmt.Errorf("set current cluster context: %w", err)
+	}
+
+	fmt.Printf("Current cluster context is now '%s'.\n", selected)
+	return nil
+}
+
+func buildContextOptions(contexts []string, current string) []huh.Option[string] {
+	options := make([]huh.Option[string], len(contexts))
+
+	for i, ctx := range contexts {
+		opt := huh.NewOption(ctx, ctx)
+		if ctx == current {
+			opt.Key += " (current)"
+			opt = opt.Selected(true)
+		}
+
+		options[i] = opt
+	}
+
+	return options
+}

cmd/uncloud/deploy.go

@@ -0,0 +1,180 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"strings"
+
+	composecli "github.com/compose-spec/compose-go/v2/cli"
+	"github.com/docker/compose/v2/pkg/progress"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/psviderski/uncloud/pkg/client"
+	"github.com/psviderski/uncloud/pkg/client/compose"
+	"github.com/psviderski/uncloud/pkg/client/deploy"
+	"github.com/spf13/cobra"
+)
+
+type deployOptions struct {
+	files    []string
+	profiles []string
+	services []string
+	noBuild  bool
+
+	context string
+}
+
+// NewDeployCommand creates a new command to deploy services from a Compose file.
+func NewDeployCommand() *cobra.Command {
+	opts := deployOptions{}
+	cmd := &cobra.Command{
+		Use:   "deploy [FLAGS] [SERVICE...]",
+		Short: "Deploy services from a Compose file.",
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+
+			if len(args) > 0 {
+				opts.services = args
+			}
+
+			return runDeploy(cmd.Context(), uncli, opts)
+		},
+	}
+
+	cmd.Flags().StringSliceVarP(&opts.files, "file", "f", nil,
+		"One or more Compose files to deploy services from. (default compose.yaml)")
+	cmd.Flags().StringSliceVarP(&opts.profiles, "profile", "p", nil,
+		"One or more Compose profiles to enable.")
+	cmd.Flags().StringVarP(&opts.context, "context", "c", "",
+		"Name of the cluster context to deploy to (default is the current context)")
+	cmd.Flags().BoolVarP(&opts.noBuild, "no-build", "n", false,
+		"Do not build images before deploying services. (default false)")
+
+	// TODO: Consider adding a filter flag to specify which machines to deploy to but keep the rest running.
+	//  Could be useful to test a new version on a subset of machines before rolling out to all.
+
+	return cmd
+}
+
+// projectOpts returns the project options for the Compose file(s).
+func projectOpts(opts deployOptions) []composecli.ProjectOptionsFn {
+	projectOpts := []composecli.ProjectOptionsFn{}
+
+	if len(opts.profiles) > 0 {
+		projectOpts = append(projectOpts, composecli.WithDefaultProfiles(opts.profiles...))
+	}
+
+	return projectOpts
+}
+
+// runDeploy parses the Compose file(s) and deploys the services.
+func runDeploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
+	projectOpts := projectOpts(opts)
+
+	project, err := compose.LoadProject(ctx, opts.files, projectOpts...)
+	if err != nil {
+		return fmt.Errorf("load compose file(s): %w", err)
+	}
+
+	if len(opts.services) > 0 {
+		// Includes service dependencies by default. This is the default docker compose behavior.
+		project, err = project.WithSelectedServices(opts.services)
+		if err != nil {
+			return fmt.Errorf("select services: %w", err)
+		}
+	}
+
+	servicesToBuild := cli.GetServicesThatNeedBuild(project)
+
+	if len(servicesToBuild) > 0 {
+		if opts.noBuild {
+			fmt.Println("Not building services as requested.")
+		} else {
+			buildOpts := cli.BuildOptions{
+				Push:    true,
+				NoCache: false,
+			}
+
+			if err := cli.BuildServices(ctx, servicesToBuild, buildOpts); err != nil {
+				return fmt.Errorf("build services: %w", err)
+			}
+		}
+	}
+
+	clusterClient, err := uncli.ConnectCluster(ctx, opts.context)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer clusterClient.Close()
+
+	composeDeploy, err := compose.NewDeployment(ctx, clusterClient, project)
+	if err != nil {
+		return fmt.Errorf("create compose deployment: %w", err)
+	}
+
+	plan, err := composeDeploy.Plan(ctx)
+	if err != nil {
+		return fmt.Errorf("plan deployment: %w", err)
+	}
+
+	if len(plan.Operations) == 0 {
+		fmt.Println("Services are up to date.")
+		return nil
+	}
+
+	fmt.Println("Deployment plan:")
+	if err = printPlan(ctx, clusterClient, plan); err != nil {
+		return fmt.Errorf("print deployment plan: %w", err)
+	}
+	fmt.Println()
+
+	confirmed, err := cli.Confirm()
+	if err != nil {
+		return fmt.Errorf("confirm deployment: %w", err)
+	}
+	if !confirmed {
+		fmt.Println("Cancelled. No changes were made.")
+		return nil
+	}
+
+	return progress.RunWithTitle(ctx, func(ctx context.Context) error {
+		if err := plan.Execute(ctx, clusterClient); err != nil {
+			return fmt.Errorf("deploy services: %w", err)
+		}
+		return nil
+	}, uncli.ProgressOut(), "Deploying services")
+}
+
+func printPlan(ctx context.Context, cli *client.Client, plan deploy.SequenceOperation) error {
+	for _, op := range plan.Operations {
+		svcPlan, ok := op.(*deploy.Plan)
+		if !ok {
+			fmt.Println("- " + op.Format(nil))
+			continue
+		}
+
+		svc, err := cli.InspectService(ctx, svcPlan.ServiceID)
+		if err != nil && !errors.Is(err, api.ErrNotFound) {
+			return fmt.Errorf("inspect service: %w", err)
+		}
+		// Initialise a machine and container name resolver to properly format the service plan output.
+		resolver, err := cli.ServiceOperationNameResolver(ctx, svc)
+		if err != nil {
+			return fmt.Errorf("create machine and container name resolver for service operations: %w", err)
+		}
+
+		fmt.Printf("- Deploy service [name=%s]\n", svcPlan.ServiceName)
+		fmt.Println(indent(svcPlan.Format(resolver), "  "))
+	}
+
+	return nil
+}
+
+func indent(text, prefix string) string {
+	lines := strings.Split(text, "\n")
+	for i, line := range lines {
+		lines[i] = prefix + line
+	}
+	return strings.Join(lines, "\n")
+}

cmd/uncloud/dns/release.go

@@ -4,15 +4,16 @@ import (
 	"context"
 	"errors"
 	"fmt"
+
+	"github.com/psviderski/uncloud/internal/cli"
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc/codes"
 	"google.golang.org/grpc/status"
 	"google.golang.org/protobuf/types/known/emptypb"
-	"uncloud/internal/cli"
 )
 
 type releaseOptions struct {
-	cluster string
+	context string
 }
 
 func NewReleaseCommand() *cobra.Command {
@@ -28,15 +29,15 @@ func NewReleaseCommand() *cobra.Command {
 	}
 
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.context, "context", "c", "",
-		"Name of the cluster. (default is the current cluster)",
+		"Name of the cluster context. (default is the current context)",
 	)
 
 	return cmd
 }
 
 func release(ctx context.Context, uncli *cli.CLI, opts releaseOptions) error {
-	client, err := uncli.ConnectCluster(ctx, opts.cluster)
+	client, err := uncli.ConnectCluster(ctx, opts.context)
 	if err != nil {
 		return fmt.Errorf("connect to cluster: %w", err)
 	}

cmd/uncloud/dns/reserve.go

@@ -4,18 +4,22 @@ import (
 	"context"
 	"errors"
 	"fmt"
+
+	"github.com/psviderski/uncloud/cmd/uncloud/caddy"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/psviderski/uncloud/pkg/client"
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc/codes"
 	"google.golang.org/grpc/status"
-	"uncloud/internal/cli"
-	"uncloud/internal/machine/api/pb"
 )
 
 const DefaultUncloudDNSAPIEndpoint = "https://dns.uncloud.run/v1"
 
 type reserveOptions struct {
 	endpoint string
-	cluster  string
+	context  string
 }
 
 func NewReserveCommand() *cobra.Command {
@@ -33,21 +37,21 @@ func NewReserveCommand() *cobra.Command {
 	cmd.Flags().StringVar(&opts.endpoint, "endpoint", DefaultUncloudDNSAPIEndpoint,
 		"API endpoint for the Uncloud DNS service.")
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.context, "context", "c", "",
-		"Name of the cluster. (default is the current cluster)",
+		"Name of the cluster context. (default is the current context)",
 	)
 
 	return cmd
 }
 
 func reserve(ctx context.Context, uncli *cli.CLI, opts reserveOptions) error {
-	client, err := uncli.ConnectCluster(ctx, opts.cluster)
+	clusterClient, err := uncli.ConnectCluster(ctx, opts.context)
 	if err != nil {
 		return fmt.Errorf("connect to cluster: %w", err)
 	}
-	defer client.Close()
+	defer clusterClient.Close()
 
-	domain, err := client.ReserveDomain(ctx, &pb.ReserveDomainRequest{Endpoint: opts.endpoint})
+	domain, err := clusterClient.ReserveDomain(ctx, &pb.ReserveDomainRequest{Endpoint: opts.endpoint})
 	if err != nil {
 		if status.Convert(err).Code() == codes.AlreadyExists {
 			return errors.New("domain already reserved")
@@ -56,7 +60,16 @@ func reserve(ctx context.Context, uncli *cli.CLI, opts reserveOptions) error {
 	}
 
 	fmt.Printf("Reserved cluster domain: %s\n", domain.Name)
-	fmt.Println("Redeploy the Caddy service ('uc caddy deploy') to configure DNS records for the domain " +
+
-		"to route traffic to the services in the cluster.")
+	// Update cluster domain records in Uncloud DNS to point to machines running caddy service if it has been deployed.
-	return nil
+	if _, err = clusterClient.InspectService(ctx, client.CaddyServiceName); err != nil {
+		if errors.Is(err, api.ErrNotFound) {
+			fmt.Println("Deploy the Caddy reverse proxy service ('uc caddy deploy') to enable internet access " +
+				"to your services via the reserved or your custom domain.")
+			return nil
+		}
+		return fmt.Errorf("inspect caddy service: %w", err)
+	}
+
+	return caddy.UpdateDomainRecords(ctx, clusterClient, uncli.ProgressOut())
 }

cmd/uncloud/dns/show.go

@@ -4,13 +4,14 @@ import (
 	"context"
 	"errors"
 	"fmt"
+
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/api"
 	"github.com/spf13/cobra"
-	"uncloud/internal/cli"
-	"uncloud/internal/cli/client"
 )
 
 type showOptions struct {
-	cluster string
+	context string
 }
 
 func NewShowCommand() *cobra.Command {
@@ -26,15 +27,15 @@ func NewShowCommand() *cobra.Command {
 	}
 
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.context, "context", "c", "",
-		"Name of the cluster. (default is the current cluster)",
+		"Name of the cluster context. (default is the current context)",
 	)
 
 	return cmd
 }
 
 func show(ctx context.Context, uncli *cli.CLI, opts showOptions) error {
-	clusterClient, err := uncli.ConnectCluster(ctx, opts.cluster)
+	clusterClient, err := uncli.ConnectCluster(ctx, opts.context)
 	if err != nil {
 		return fmt.Errorf("connect to cluster: %w", err)
 	}
@@ -42,7 +43,7 @@ func show(ctx context.Context, uncli *cli.CLI, opts showOptions) error {
 
 	domain, err := clusterClient.GetDomain(ctx)
 	if err != nil {
-		if errors.Is(err, client.ErrNotFound) {
+		if errors.Is(err, api.ErrNotFound) {
 			return errors.New("no domain reserved")
 		}
 		return err

cmd/uncloud/machine/add.go

@@ -4,19 +4,19 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"net/netip"
+	"time"
+
 	"github.com/cenkalti/backoff/v4"
 	"github.com/docker/compose/v2/pkg/progress"
+	"github.com/psviderski/uncloud/cmd/uncloud/caddy"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/internal/cli/config"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/psviderski/uncloud/pkg/client"
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc/codes"
 	"google.golang.org/grpc/status"
-	"google.golang.org/protobuf/types/known/emptypb"
-	"net/netip"
-	"time"
-	"uncloud/cmd/uncloud/caddy"
-	"uncloud/internal/cli"
-	"uncloud/internal/cli/client"
-	"uncloud/internal/cli/config"
-	"uncloud/internal/machine/api/pb"
 )
 
 type addOptions struct {
@@ -24,7 +24,8 @@ type addOptions struct {
 	noCaddy  bool
 	publicIP string
 	sshKey   string
-	cluster  string
+	context  string
+	version  string
 }
 
 func NewAddCommand() *cobra.Command {
@@ -61,13 +62,18 @@ func NewAddCommand() *cobra.Command {
 			"blank '' or 'none' to disable ingress on this machine, or specify an IP address.",
 	)
 	cmd.Flags().StringVarP(
-		&opts.sshKey, "ssh-key", "i", "",
+		&opts.sshKey, "ssh-key", "i", "~/.ssh/id_ed25519",
-		"path to SSH private key for SSH remote login. (default ~/.ssh/id_*)",
+		"Path to SSH private key for remote login (if not already added to SSH agent).",
+	)
+	cmd.Flags().StringVar(
+		&opts.version, "version", "latest",
+		"Version of the Uncloud daemon to install on the machine.",
 	)
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.context, "context", "c", "",
-		"Name of the cluster to add the machine to. (default is the current cluster)",
+		"Name of the cluster context to add the machine to. (default is the current context)",
 	)
+
 	return cmd
 }
 
@@ -86,10 +92,17 @@ func add(ctx context.Context, uncli *cli.CLI, remoteMachine cli.RemoteMachine, o
 		publicIP = &ip
 	}
 
-	machineClient, err := uncli.AddMachine(ctx, remoteMachine, opts.cluster, opts.name, publicIP)
+	clusterClient, machineClient, err := uncli.AddMachine(ctx, cli.AddMachineOptions{
+		Context:       opts.context,
+		MachineName:   opts.name,
+		PublicIP:      publicIP,
+		RemoteMachine: remoteMachine,
+		Version:       opts.version,
+	})
 	if err != nil {
 		return err
 	}
+	defer clusterClient.Close()
 	defer machineClient.Close()
 
 	if opts.noCaddy {
@@ -98,24 +111,20 @@ func add(ctx context.Context, uncli *cli.CLI, remoteMachine cli.RemoteMachine, o
 
 	// Wait for the cluster to be initialised to be able to deploy the Caddy service.
 	fmt.Println("Waiting for the machine to be ready...")
+	fmt.Println()
 	if err = waitClusterInitialised(ctx, machineClient); err != nil {
 		return fmt.Errorf("wait for cluster to be initialised on machine: %w", err)
 	}
 
-	// Inspect the added machine to get its ID to create a filter for the Caddy deployment.
-	minfo, err := machineClient.Inspect(ctx, &emptypb.Empty{})
-	if err != nil {
-		return fmt.Errorf("inspect machine: %w", err)
-	}
-	filter := func(m *pb.MachineInfo) bool {
-		return m.Id == minfo.Id
-	}
 	// Deploy a Caddy service container to the added machine. If caddy service is already deployed on other machines,
 	// use the deployed image version. Otherwise, use the latest version.
+	// NOTE: We use the cluster client to inspect and scale the Caddy service because the newly added machine may have
+	// issues accessing the Machine API of existing machines in the cluster.
+	// See the issue for more details: https://github.com/psviderski/uncloud/issues/65.
 	caddyImage := ""
-	caddySvc, err := machineClient.InspectService(ctx, client.CaddyServiceName)
+	caddySvc, err := clusterClient.InspectService(ctx, client.CaddyServiceName)
 	if err != nil {
-		if !errors.Is(err, client.ErrNotFound) {
+		if !errors.Is(err, api.ErrNotFound) {
 			return fmt.Errorf("inspect caddy service: %w", err)
 		}
 	} else {
@@ -134,7 +143,9 @@ func add(ctx context.Context, uncli *cli.CLI, remoteMachine cli.RemoteMachine, o
 		}
 	}
 
-	d, err := machineClient.NewCaddyDeployment(caddyImage, filter)
+	// TODO: scale the existing Caddy service to the new machine instead of running a new deployment
+	//  that may cause a small downtime.
+	d, err := clusterClient.NewCaddyDeployment(caddyImage, api.Placement{})
 	if err != nil {
 		return fmt.Errorf("create caddy deployment: %w", err)
 	}
@@ -160,7 +171,7 @@ func waitClusterInitialised(ctx context.Context, client *client.Client) error {
 	), ctx)
 
 	check := func() error {
-		_, err := client.ListMachines(ctx)
+		_, err := client.ListMachines(ctx, nil)
 		if err == nil {
 			return nil
 		}

cmd/uncloud/machine/init.go

@@ -3,15 +3,17 @@ package machine
 import (
 	"context"
 	"fmt"
-	"github.com/docker/compose/v2/pkg/progress"
-	"github.com/spf13/cobra"
 	"net/netip"
-	"uncloud/cmd/uncloud/caddy"
+
-	"uncloud/cmd/uncloud/dns"
+	"github.com/docker/compose/v2/pkg/progress"
-	"uncloud/internal/cli"
+	"github.com/psviderski/uncloud/cmd/uncloud/caddy"
-	"uncloud/internal/cli/config"
+	"github.com/psviderski/uncloud/cmd/uncloud/dns"
-	"uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/internal/cli"
-	"uncloud/internal/machine/cluster"
+	"github.com/psviderski/uncloud/internal/cli/config"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/internal/machine/cluster"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/spf13/cobra"
 )
 
 type initOptions struct {
@@ -22,15 +24,17 @@ type initOptions struct {
 	noDNS       bool
 	publicIP    string
 	sshKey      string
-	cluster     string
+	version     string
+	context     string
 }
 
 func NewInitCommand() *cobra.Command {
 	opts := initOptions{}
 	cmd := &cobra.Command{
 		Use:   "init [USER@HOST:PORT]",
-		Short: "Initialise a new cluster that consists of the local or remote machine.",
+		Short: "Initialise a new cluster with a remote machine as the first member.",
-		// TODO: include usage examples of initialising a local and remote machine.
+		// TODO: include usage examples of initialising a remote machine.
+		// TODO: support initialising a cluster on the local machine.
 		Args: cobra.MaximumNArgs(1),
 		RunE: func(cmd *cobra.Command, args []string) error {
 			uncli := cmd.Context().Value("cli").(*cli.CLI)
@@ -76,12 +80,16 @@ func NewInitCommand() *cobra.Command {
 			"blank '' or 'none' to disable ingress on this machine, or specify an IP address.",
 	)
 	cmd.Flags().StringVarP(
-		&opts.sshKey, "ssh-key", "i", "",
+		&opts.sshKey, "ssh-key", "i", "~/.ssh/id_ed25519",
-		"path to SSH private key for SSH remote login. (default ~/.ssh/id_*)",
+		"Path to SSH private key for remote login (if not already added to SSH agent).",
+	)
+	cmd.Flags().StringVar(
+		&opts.version, "version", "latest",
+		"Version of the Uncloud daemon to install on the machine.",
 	)
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.context, "context", "c", "default",
-		"Name of the cluster in the local config if initialising a remote machine.",
+		"Name of the created context for the initialised cluster in the Uncloud config.",
 	)
 
 	return cmd
@@ -106,8 +114,14 @@ func initCluster(ctx context.Context, uncli *cli.CLI, remoteMachine *cli.RemoteM
 		}
 		publicIP = &ip
 	}
-
+	client, err := uncli.InitCluster(ctx, cli.InitClusterOptions{
-	client, err := uncli.InitCluster(ctx, remoteMachine, opts.cluster, opts.name, netPrefix, publicIP)
+		Context:       opts.context,
+		MachineName:   opts.name,
+		Network:       netPrefix,
+		PublicIP:      publicIP,
+		RemoteMachine: remoteMachine,
+		Version:       opts.version,
+	})
 	if err != nil {
 		return err
 	}
@@ -121,6 +135,7 @@ func initCluster(ctx context.Context, uncli *cli.CLI, remoteMachine *cli.RemoteM
 	// The creation of a deployment plan talks to cluster API. Since the API needs a few moments to become available
 	// after cluster initialisation, we keep the user informed during this wait.
 	fmt.Println("Waiting for the machine to be ready...")
+	fmt.Println()
 
 	if !opts.noDNS {
 		domain, err := client.ReserveDomain(ctx, &pb.ReserveDomainRequest{Endpoint: opts.dnsEndpoint})
@@ -131,7 +146,7 @@ func initCluster(ctx context.Context, uncli *cli.CLI, remoteMachine *cli.RemoteM
 	}
 
 	if !opts.noCaddy {
-		d, err := client.NewCaddyDeployment("", nil)
+		d, err := client.NewCaddyDeployment("", api.Placement{})
 		if err != nil {
 			return fmt.Errorf("create caddy deployment: %w", err)
 		}

cmd/uncloud/machine/ls.go

@@ -3,29 +3,30 @@ package machine
 import (
 	"context"
 	"fmt"
-	"github.com/spf13/cobra"
 	"net/netip"
 	"os"
 	"strings"
 	"text/tabwriter"
-	"uncloud/internal/cli"
+
-	"uncloud/internal/machine/network"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/internal/machine/network"
+	"github.com/spf13/cobra"
 )
 
 func NewListCommand() *cobra.Command {
-	var cluster string
+	var contextName string
 	cmd := &cobra.Command{
 		Use:     "ls",
 		Aliases: []string{"list"},
 		Short:   "List machines in a cluster.",
 		RunE: func(cmd *cobra.Command, args []string) error {
 			uncli := cmd.Context().Value("cli").(*cli.CLI)
-			return list(cmd.Context(), uncli, cluster)
+			return list(cmd.Context(), uncli, contextName)
 		},
 	}
 	cmd.Flags().StringVarP(
-		&cluster, "cluster", "c", "",
+		&contextName, "context", "c", "",
-		"Name of the cluster. (default is the current cluster)",
+		"Name of the cluster context. (default is the current context)",
 	)
 	return cmd
 }
@@ -37,7 +38,7 @@ func list(ctx context.Context, uncli *cli.CLI, clusterName string) error {
 	}
 	defer client.Close()
 
-	machines, err := client.ListMachines(ctx)
+	machines, err := client.ListMachines(ctx, nil)
 	if err != nil {
 		return fmt.Errorf("list machines: %w", err)
 	}
@@ -67,7 +68,8 @@ func list(ctx context.Context, uncli *cli.CLI, clusterName string) error {
 		}
 
 		if _, err = fmt.Fprintf(
-			tw, "%s\t%s\t%s\t%s\t%s\n", m.Name, capitalise(member.State.String()), subnet, publicIP, strings.Join(endpoints, ", "),
+			tw, "%s\t%s\t%s\t%s\t%s\n", m.Name, capitalise(member.State.String()), subnet, publicIP,
+			strings.Join(endpoints, ", "),
 		); err != nil {
 			return fmt.Errorf("write row: %w", err)
 		}

cmd/uncloud/machine/rm.go

@@ -0,0 +1,139 @@
+package machine
+
+import (
+	"context"
+	"fmt"
+	"maps"
+	"slices"
+	"strings"
+
+	"github.com/charmbracelet/lipgloss"
+	"github.com/charmbracelet/lipgloss/tree"
+	"github.com/docker/docker/api/types/container"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/spf13/cobra"
+)
+
+type removeOptions struct {
+	force   bool
+	yes     bool
+	context string
+}
+
+func NewRmCommand() *cobra.Command {
+	opts := removeOptions{}
+
+	cmd := &cobra.Command{
+		Use:     "rm MACHINE",
+		Aliases: []string{"remove", "delete"},
+		Short:   "Remove a machine from a cluster.",
+		Args:    cobra.ExactArgs(1),
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+			return remove(cmd.Context(), uncli, args[0], opts)
+		},
+	}
+
+	cmd.Flags().StringVarP(&opts.context, "context", "c", "",
+		"Name of the cluster context. (default is the current context)")
+	cmd.Flags().BoolVarP(&opts.yes, "yes", "y", false,
+		"Do not prompt for confirmation before removing the machine.")
+
+	return cmd
+}
+
+func remove(ctx context.Context, uncli *cli.CLI, machineName string, opts removeOptions) error {
+	client, err := uncli.ConnectCluster(ctx, opts.context)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer client.Close()
+
+	// Verify the machine exists and list all service containers on it including stopped ones.
+	listCtx, machines, err := api.ProxyMachinesContext(ctx, client, []string{machineName})
+	if err != nil {
+		return err
+	}
+	if len(machines) == 0 {
+		return fmt.Errorf("machine '%s' not found in the cluster", machineName)
+	}
+	m := machines[0].Machine
+
+	listOpts := container.ListOptions{All: true}
+	machineContainers, err := client.Docker.ListServiceContainers(listCtx, "", listOpts)
+	if err != nil {
+		return fmt.Errorf("list containers: %w", err)
+	}
+	containers := machineContainers[0].Containers
+
+	if len(containers) > 0 {
+		fmt.Printf("Found %d service containers on machine '%s':\n\n", len(containers), m.Name)
+		fmt.Println(formatContainerTree(containers))
+		fmt.Println()
+		fmt.Println("This will remove all service containers on the machine, reset it to the uninitialised state, " +
+			"and remove it from the cluster.")
+	} else {
+		fmt.Printf("No service containers found on machine '%s'.\n", m.Name)
+		fmt.Println("This will reset the machine to the uninitialised state and remove it from the cluster.")
+	}
+
+	if !opts.yes {
+		confirmed, err := cli.Confirm()
+		if err != nil {
+			return fmt.Errorf("confirm removal: %w", err)
+		}
+		if !confirmed {
+			fmt.Println("Cancelled. Machine was not removed.")
+			return nil
+		}
+	}
+
+	// TODO: 3. Remove all service containers on the machine.
+	// TODO: 4. Implement and call ResetMachine via Machine API to reset the machine state to uninitialised.
+	// TODO: 5. Remove the machine from the cluster store.
+
+	fmt.Printf("Machine '%s' removed from the cluster.\n", m.Name)
+	return nil
+}
+
+// formatContainerTree formats a list of containers grouped by service as a tree structure.
+func formatContainerTree(containers []api.ServiceContainer) string {
+	if len(containers) == 0 {
+		return ""
+	}
+
+	// Group containers by service.
+	serviceContainers := make(map[string][]api.ServiceContainer)
+	for _, ctr := range containers {
+		serviceName := ctr.ServiceName()
+		serviceContainers[serviceName] = append(serviceContainers[serviceName], ctr)
+	}
+
+	// Build tree output.
+	var output []string
+	serviceNames := slices.Sorted(maps.Keys(serviceContainers))
+	for _, serviceName := range serviceNames {
+		ctrs := serviceContainers[serviceName]
+		mode := ctrs[0].ServiceMode()
+
+		// Format a tree for the service with its containers.
+		plural := ""
+		if len(ctrs) > 1 {
+			plural = "s"
+		}
+		t := tree.Root(fmt.Sprintf("• %s (%s, %d container%s)", serviceName, mode, len(ctrs), plural)).
+			EnumeratorStyle(lipgloss.NewStyle().MarginLeft(2).MarginRight(1))
+
+		// Add containers as children.
+		for _, ctr := range ctrs {
+			state, _ := ctr.HumanState()
+			info := fmt.Sprintf("%s • %s • %s", ctr.Name, ctr.Config.Image, state)
+			t.Child(info)
+		}
+
+		output = append(output, t.String())
+	}
+
+	return strings.Join(output, "\n")
+}

cmd/uncloud/machine/root.go

@@ -6,13 +6,15 @@ import (
 
 func NewRootCommand() *cobra.Command {
 	cmd := &cobra.Command{
-		Use:   "machine",
+		Use:     "machine",
-		Short: "Manage machines in an Uncloud cluster.",
+		Aliases: []string{"m"},
+		Short:   "Manage machines in an Uncloud cluster.",
 	}
 	cmd.AddCommand(
 		NewAddCommand(),
 		NewInitCommand(),
 		NewListCommand(),
+		NewRmCommand(),
 		NewTokenCommand(),
 	)
 	return cmd

cmd/uncloud/machine/token.go

@@ -3,8 +3,8 @@ package machine
 import (
 	"fmt"
 	"github.com/spf13/cobra"
-	"uncloud/internal/daemon"
+	"github.com/psviderski/uncloud/internal/daemon"
-	"uncloud/internal/machine"
+	"github.com/psviderski/uncloud/internal/machine"
 )
 
 type tokenOptions struct {

cmd/uncloud/main.go

@@ -3,47 +3,81 @@ package main
 import (
 	"context"
 	"fmt"
-	"github.com/spf13/cobra"
+	"net/netip"
-	"os"
 	"strings"
-	"uncloud/cmd/uncloud/caddy"
+
-	"uncloud/cmd/uncloud/dns"
+	"github.com/psviderski/uncloud/cmd/uncloud/caddy"
-	"uncloud/cmd/uncloud/machine"
+	cmdcontext "github.com/psviderski/uncloud/cmd/uncloud/context"
-	"uncloud/cmd/uncloud/service"
+	"github.com/psviderski/uncloud/cmd/uncloud/dns"
-	"uncloud/internal/cli"
+	"github.com/psviderski/uncloud/cmd/uncloud/machine"
+	"github.com/psviderski/uncloud/cmd/uncloud/service"
+	"github.com/psviderski/uncloud/cmd/uncloud/volume"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/internal/cli/config"
+	"github.com/psviderski/uncloud/internal/fs"
+	"github.com/psviderski/uncloud/internal/version"
+	"github.com/spf13/cobra"
 )
 
+type globalOptions struct {
+	configPath string
+	connect    string
+}
+
 func main() {
-	var configPath string
+	opts := globalOptions{}
 	cmd := &cobra.Command{
 		Use:           "uncloud",
 		Short:         "A CLI tool for managing Uncloud resources such as clusters, machines, and services.",
+		Version:       version.String(),
 		SilenceUsage:  true,
 		SilenceErrors: true,
 		PersistentPreRunE: func(cmd *cobra.Command, args []string) error {
-			if strings.HasPrefix(configPath, "~/") {
+			var conn *config.MachineConnection
-				home, err := os.UserHomeDir()
+			if opts.connect != "" {
-				if err != nil {
+				if strings.HasPrefix(opts.connect, "tcp://") {
-					return fmt.Errorf("get user home directory to resolve %q: %w", configPath, err)
+					addrPort, err := netip.ParseAddrPort(opts.connect[len("tcp://"):])
+					if err != nil {
+						return fmt.Errorf("parse TCP address: %w", err)
+					}
+					conn = &config.MachineConnection{
+						TCP: &addrPort,
+					}
+				} else {
+					dest := opts.connect
+					if strings.HasPrefix(dest, "ssh://") {
+						dest = dest[len("ssh://"):]
+					}
+					conn = &config.MachineConnection{
+						SSH: config.SSHDestination(dest),
+					}
 				}
-				configPath = strings.Replace(configPath, "~", home, 1)
 			}
 
-			uncli, err := cli.New(configPath)
+			configPath := fs.ExpandHomeDir(opts.configPath)
+			uncli, err := cli.New(configPath, conn)
 			if err != nil {
-				return fmt.Errorf("initialize CLI: %w", err)
+				return fmt.Errorf("initialise CLI: %w", err)
 			}
 			cmd.SetContext(context.WithValue(cmd.Context(), "cli", uncli))
 			return nil
 		},
 	}
+
+	cmd.PersistentFlags().StringVar(&opts.connect, "connect", "",
+		"Connect to a remote cluster machine without using the Uncloud configuration file.\n"+
+			"Format: [ssh://]user@host[:port] or tcp://host:port")
 	// TODO: allow to override using UNCLOUD_CONFIG env var.
-	cmd.PersistentFlags().StringVar(&configPath, "uncloud-config", "~/.config/uncloud/config.toml",
+	cmd.PersistentFlags().StringVar(&opts.configPath, "uncloud-config", "~/.config/uncloud/config.yaml",
-		"path to the Uncloud configuration file.")
+		"Path to the Uncloud configuration file.")
-	_ = cmd.MarkPersistentFlagFilename("uncloud-config", "toml")
+	_ = cmd.MarkPersistentFlagFilename("uncloud-config", "yaml", "yml")
+	// TODO: make --context a global flag and pass it as a value of the command context.
 
 	cmd.AddCommand(
+		NewDeployCommand(),
+		NewBuildCommand(),
 		caddy.NewRootCommand(),
+		cmdcontext.NewRootCommand(),
 		dns.NewRootCommand(),
 		machine.NewRootCommand(),
 		service.NewRootCommand(),
@@ -51,6 +85,8 @@ func main() {
 		service.NewListCommand(),
 		service.NewRmCommand(),
 		service.NewRunCommand(),
+		service.NewScaleCommand(),
+		volume.NewRootCommand(),
 	)
 	cobra.CheckErr(cmd.Execute())
 }

cmd/uncloud/service/inspect.go

@@ -3,14 +3,15 @@ package service
 import (
 	"context"
 	"fmt"
-	"github.com/docker/docker/pkg/stringid"
-	"github.com/docker/go-units"
 	"os"
 	"text/tabwriter"
 	"time"
 
+	"github.com/docker/docker/pkg/stringid"
+	"github.com/docker/go-units"
+
+	"github.com/psviderski/uncloud/internal/cli"
 	"github.com/spf13/cobra"
-	"uncloud/internal/cli"
 )
 
 type inspectOptions struct {
@@ -21,7 +22,7 @@ type inspectOptions struct {
 func NewInspectCommand() *cobra.Command {
 	opts := inspectOptions{}
 	cmd := &cobra.Command{
-		Use:   "inspect",
+		Use:   "inspect SERVICE",
 		Short: "Display detailed information on a service.",
 		Args:  cobra.ExactArgs(1),
 		RunE: func(cmd *cobra.Command, args []string) error {
@@ -31,8 +32,8 @@ func NewInspectCommand() *cobra.Command {
 		},
 	}
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.cluster, "context", "c", "",
-		"Name of the cluster. (default is the current cluster)",
+		"Name of the cluster context. (default is the current context)",
 	)
 	return cmd
 }
@@ -49,7 +50,7 @@ func inspect(ctx context.Context, uncli *cli.CLI, opts inspectOptions) error {
 		return fmt.Errorf("inspect service: %w", err)
 	}
 
-	machines, err := client.ListMachines(ctx)
+	machines, err := client.ListMachines(ctx, nil)
 	if err != nil {
 		return fmt.Errorf("list machines: %w", err)
 	}

cmd/uncloud/service/ls.go

@@ -3,14 +3,16 @@ package service
 import (
 	"context"
 	"fmt"
-	"github.com/spf13/cobra"
 	"os"
 	"strings"
 	"text/tabwriter"
-	"uncloud/internal/cli"
+
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/spf13/cobra"
 )
 
 func NewListCommand() *cobra.Command {
+	// TODO(lhf): rename to context
 	var cluster string
 	cmd := &cobra.Command{
 		Use:     "ls",
@@ -22,8 +24,8 @@ func NewListCommand() *cobra.Command {
 		},
 	}
 	cmd.Flags().StringVarP(
-		&cluster, "cluster", "c", "",
+		&cluster, "context", "c", "",
-		"Name of the cluster. (default is the current cluster)",
+		"Name of the cluster context. (default is the current context)",
 	)
 	return cmd
 }

cmd/uncloud/service/rm.go

@@ -3,11 +3,10 @@ package service
 import (
 	"context"
 	"fmt"
-	"github.com/docker/cli/cli/streams"
+
 	"github.com/docker/compose/v2/pkg/progress"
+	"github.com/psviderski/uncloud/internal/cli"
 	"github.com/spf13/cobra"
-	"os"
-	"uncloud/internal/cli"
 )
 
 type rmOptions struct {
@@ -29,8 +28,8 @@ func NewRmCommand() *cobra.Command {
 		},
 	}
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.cluster, "context", "c", "",
-		"Name of the cluster. (default is the current cluster)",
+		"Name of the cluster context. (default is the current context)",
 	)
 	return cmd
 }
@@ -48,7 +47,7 @@ func rm(ctx context.Context, uncli *cli.CLI, opts rmOptions) error {
 				return fmt.Errorf("remove service '%s': %w", s, err)
 			}
 			return nil
-		}, streams.NewOut(os.Stdout), "Removing service "+s)
+		}, uncli.ProgressOut(), "Removing service "+s)
 	}
 
 	return nil

cmd/uncloud/service/root.go

@@ -6,14 +6,16 @@ import (
 
 func NewRootCommand() *cobra.Command {
 	cmd := &cobra.Command{
-		Use:   "service",
+		Use:     "service",
-		Short: "Manage services in an Uncloud cluster.",
+		Aliases: []string{"svc"},
+		Short:   "Manage services in an Uncloud cluster.",
 	}
 	cmd.AddCommand(
 		NewInspectCommand(),
 		NewListCommand(),
 		NewRmCommand(),
 		NewRunCommand(),
+		NewScaleCommand(),
 	)
 	return cmd
 }

cmd/uncloud/service/run.go

@@ -3,24 +3,36 @@ package service
 import (
 	"context"
 	"fmt"
-	"github.com/spf13/cobra"
+	"os"
-	"slices"
 	"strings"
-	"uncloud/internal/api"
+
-	"uncloud/internal/cli"
+	dockeropts "github.com/docker/cli/opts"
-	"uncloud/internal/cli/client"
+	"github.com/docker/compose/v2/pkg/progress"
-	"uncloud/internal/machine/api/pb"
+	"github.com/docker/docker/daemon/names"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/internal/secret"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/psviderski/uncloud/pkg/client/deploy"
+	"github.com/spf13/cobra"
 )
 
 type runOptions struct {
-	command  []string
+	command           []string
-	image    string
+	cpu               dockeropts.NanoCPUs
-	machines []string
+	entrypoint        string
-	mode     string
+	entrypointChanged bool
-	name     string
+	env               []string
-	publish  []string
+	image             string
-	replicas uint
+	machines          []string
-	volumes  []string
+	memory            dockeropts.MemBytes
+	mode              string
+	name              string
+	privileged        bool
+	publish           []string
+	pull              string
+	replicas          uint
+	user              string
+	volumes           []string
 
 	cluster string
 }
@@ -35,6 +47,7 @@ func NewRunCommand() *cobra.Command {
 		RunE: func(cmd *cobra.Command, args []string) error {
 			uncli := cmd.Context().Value("cli").(*cli.CLI)
 
+			opts.entrypointChanged = cmd.Flag("entrypoint").Changed
 			opts.image = args[0]
 			if len(args) > 1 {
 				opts.command = args[1:]
@@ -44,115 +57,307 @@ func NewRunCommand() *cobra.Command {
 		},
 	}
 
+	cmd.Flags().VarP(&opts.cpu, "cpu", "",
+		"Maximum number of CPU cores a service container can use. Fractional values are allowed: "+
+			"0.5 for half a core or 2.25 for two and a quarter cores.")
+	cmd.Flags().StringVar(&opts.entrypoint, "entrypoint", "",
+		"Overwrite the default ENTRYPOINT of the image. Pass an empty string \"\" to reset it.")
+	cmd.Flags().StringSliceVarP(&opts.env, "env", "e", nil,
+		"Set an environment variable for service containers. Can be specified multiple times.\n"+
+			"Format: VAR=value or just VAR to use the value from the local environment.")
 	cmd.Flags().StringVar(&opts.mode, "mode", api.ServiceModeReplicated,
-		fmt.Sprintf("Replication mode of the service: either %q (a specified number of containers across "+
+		fmt.Sprintf("Replication mode of the service: either '%s' (a specified number of containers across "+
-			"the machines) or %q (one container on every machine).",
+			"the machines) or '%s' (one container on every machine).",
 			api.ServiceModeReplicated, api.ServiceModeGlobal))
 	cmd.Flags().StringSliceVarP(&opts.machines, "machine", "m", nil,
-		"Placement constraint by machine name, limiting which machines the service can run on. Can be specified "+
+		"Placement constraint by machine names, limiting which machines the service can run on. Can be specified "+
 			"multiple times or as a comma-separated list of machine names. (default is any suitable machine)")
+	cmd.Flags().VarP(&opts.memory, "memory", "",
+		"Maximum amount of memory a service container can use. Value is a positive integer with optional unit suffix "+
+			"(b, k, m, g). Default unit is bytes if no suffix specified.\n"+
+			"Examples: 1073741824, 1024m, 1g (all equal 1 gibibyte)")
 	cmd.Flags().StringVarP(&opts.name, "name", "n", "",
 		"Assign a name to the service. A random name is generated if not specified.")
+	cmd.Flags().BoolVar(&opts.privileged, "privileged", false,
+		"Give extended privileges to service containers. This is a security risk and should be used with caution.")
 	cmd.Flags().StringSliceVarP(&opts.publish, "publish", "p", nil,
 		"Publish a service port to make it accessible outside the cluster. Can be specified multiple times.\n"+
 			"Format: [hostname:][load_balancer_port:]container_port[/protocol] or [host_ip:]:host_port:container_port[/protocol]@host\n"+
-			"Supported protocols: tcp, udp, http, https (default is tcp). If a hostname for http(s) port is not specified,\n"+
+			"Supported protocols: tcp, udp, http, https (default is tcp). If a hostname for http(s) port is not specified\n"+
-			"service-name.cluster-domain will be used as the hostname.\n"+
+			"and a cluster domain is reserved, service-name.cluster-domain will be used as the hostname.\n"+
 			"Examples:\n"+
-			"  -p 8080/https                  Publish port 8080 as HTTPS via load balancer with default service-name.cluster-domain hostname\n"+
+			"  -p 8080/https                  Publish port 8080 as HTTPS via reverse proxy with default service-name.cluster-domain hostname\n"+
-			"  -p app.example.com:8080/https  Publish port 8080 as HTTPS via load balancer with custom hostname\n"+
+			"  -p app.example.com:8080/https  Publish port 8080 as HTTPS via reverse proxy with custom hostname\n"+
-			"  -p 9000:8080                   Publish port 8080 as TCP port 9000 via load balancer\n"+
+			// TODO: add support for publishing L4 tcp/udp ports.
+			//"  -p 9000:8080                   Publish port 8080 as TCP port 9000 via reverse proxy\n"+
 			"  -p 53:5353/udp@host            Bind UDP port 5353 to host port 53")
+	cmd.Flags().StringVar(&opts.pull, "pull", api.PullPolicyMissing,
+		fmt.Sprintf("Pull image from the registry before running service containers ('%s', '%s', '%s').",
+			api.PullPolicyAlways, api.PullPolicyMissing, api.PullPolicyNever))
 	cmd.Flags().UintVar(&opts.replicas, "replicas", 1,
 		"Number of containers to run for the service. Only valid for a replicated service.")
+	cmd.Flags().StringVarP(&opts.user, "user", "u", "",
+		"User name or UID and optionally group name or GID used for running the command inside service containers.\n"+
+			"Format: USER[:GROUP] or UID[:GID]. If not specified, the user is set to the default user of the image.")
 	cmd.Flags().StringSliceVarP(&opts.volumes, "volume", "v", nil,
-		"Bind mount a host file or directory into a service container using the format "+
+		"Mount a data volume or host path into service containers. Service containers will be scheduled on the machine(s) where\n"+
-			"/host/path:/container/path[:ro]. Can be specified multiple times.")
+			"the volume is located. Can be specified multiple times.\n"+
+			"Format: volume_name:/container/path[:ro|volume-nocopy] or /host/path:/container/path[:ro]\n"+
+			"Examples:\n"+
+			"  -v postgres-data:/var/lib/postgresql/data  Mount volume 'postgres-data' to /var/lib/postgresql/data in container\n"+
+			"  -v /data/uploads:/app/uploads         	 Bind mount /data/uploads host directory to /app/uploads in container\n"+
+			"  -v /host/path:/container/path:ro 		 Bind mount a host directory or file as read-only")
 
 	cmd.Flags().StringVarP(
-		&opts.cluster, "cluster", "c", "",
+		&opts.cluster, "context", "c", "",
-		"Name of the cluster to run the service in. (default is the current cluster)",
+		"Name of the cluster context to run the service in. (default is the current context)",
 	)
 
 	return cmd
 }
 
 func run(ctx context.Context, uncli *cli.CLI, opts runOptions) error {
+	spec, err := prepareServiceSpec(opts)
+	if err != nil {
+		return err
+	}
+
+	clusterClient, err := uncli.ConnectCluster(ctx, opts.cluster)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer clusterClient.Close()
+
+	var resp api.RunServiceResponse
+	err = progress.RunWithTitle(ctx, func(ctx context.Context) error {
+		resp, err = clusterClient.RunService(ctx, spec)
+		if err != nil {
+			return fmt.Errorf("run service: %w", err)
+		}
+
+		return nil
+	}, uncli.ProgressOut(), fmt.Sprintf("Running service %s (%s mode)", spec.Name, spec.Mode))
+	if err != nil {
+		return err
+	}
+
+	svc, err := clusterClient.InspectService(ctx, resp.ID)
+	if err != nil {
+		return fmt.Errorf("inspect service: %w", err)
+	}
+
+	endpoints := svc.Endpoints()
+	if len(endpoints) > 0 {
+		fmt.Println()
+		fmt.Printf("%s endpoints:\n", svc.Name)
+		for _, endpoint := range endpoints {
+			fmt.Printf(" • %s\n", endpoint)
+		}
+	}
+
+	return nil
+}
+
+func prepareServiceSpec(opts runOptions) (api.ServiceSpec, error) {
+	var spec api.ServiceSpec
+
+	env, err := parseEnv(opts.env)
+	if err != nil {
+		return spec, err
+	}
+
 	switch opts.mode {
 	case api.ServiceModeReplicated, api.ServiceModeGlobal:
 	default:
-		return fmt.Errorf("invalid replication mode: %q", opts.mode)
+		return spec, fmt.Errorf("invalid replication mode: '%s'", opts.mode)
 	}
 
-	var machineFilter client.MachineFilter
+	switch opts.pull {
-	if len(opts.machines) > 0 {
+	case api.PullPolicyAlways, api.PullPolicyMissing, api.PullPolicyNever:
-		var machines []string
+	default:
-		for _, value := range opts.machines {
+		return spec, fmt.Errorf("invalid pull policy: '%s'", opts.pull)
-			if value == "" {
-				continue
-			}
-
-			mlist := strings.Split(value, ",")
-			for _, m := range mlist {
-				if m = strings.TrimSpace(m); m != "" {
-					machines = append(machines, m)
-				}
-			}
-		}
-
-		if len(machines) > 0 {
-			machineFilter = func(m *pb.MachineInfo) bool {
-				return slices.Contains(machines, m.Name)
-			}
-		}
 	}
 
 	ports := make([]api.PortSpec, len(opts.publish))
 	for i, publishPort := range opts.publish {
 		port, err := api.ParsePortSpec(publishPort)
 		if err != nil {
-			return fmt.Errorf("invalid service port '%s': %w", publishPort, err)
+			return spec, fmt.Errorf("invalid service port '%s': %w", publishPort, err)
 		}
 		ports[i] = port
 	}
-	// TODO: parse and validate opts.volumes to fail fast if invalid.
 
-	spec := api.ServiceSpec{
+	volumes, mounts, err := parseVolumeFlags(opts.volumes)
+	if err != nil {
+		return spec, err
+	}
+
+	placement := api.Placement{
+		Machines: cli.ExpandCommaSeparatedValues(opts.machines),
+	}
+
+	spec = api.ServiceSpec{
 		Container: api.ContainerSpec{
-			Command: opts.command,
+			Command:    opts.command,
-			Image:   opts.image,
+			Env:        env,
-			Volumes: opts.volumes,
+			Image:      opts.image,
+			Privileged: opts.privileged,
+			PullPolicy: opts.pull,
+			Resources: api.ContainerResources{
+				CPU:    opts.cpu.Value(),
+				Memory: opts.memory.Value(),
+			},
+			User:         opts.user,
+			VolumeMounts: mounts,
 		},
-		Mode:     opts.mode,
+		Mode:      opts.mode,
-		Name:     opts.name,
+		Name:      opts.name,
-		Ports:    ports,
+		Placement: placement,
-		Replicas: opts.replicas,
+		Ports:     ports,
-	}
+		Replicas:  opts.replicas,
-	if err := spec.Validate(); err != nil {
+		Volumes:   volumes,
-		return fmt.Errorf("invalid service configuration: %w", err)
 	}
 
-	client, err := uncli.ConnectCluster(ctx, opts.cluster)
+	// Overwrite the default ENTRYPOINT of the image or reset it if an empty string is passed.
-	if err != nil {
+	if opts.entrypoint != "" {
-		return fmt.Errorf("connect to cluster: %w", err)
+		spec.Container.Entrypoint = []string{opts.entrypoint}
-	}
+	} else if opts.entrypointChanged {
-	defer client.Close()
+		spec.Container.Entrypoint = []string{""}
-
-	resp, err := client.RunService(ctx, spec, machineFilter)
-	if err != nil {
-		return fmt.Errorf("run service: %w", err)
 	}
 
-	svc, err := client.InspectService(ctx, resp.ID)
+	if err = spec.Validate(); err != nil {
-	if err != nil {
+		return spec, fmt.Errorf("invalid service configuration: %w", err)
-		return fmt.Errorf("inspect service: %w", err)
 	}
 
-	fmt.Println()
+	// Generate a service name if not specified to be able to include it in the progress title.
-	fmt.Printf("%s endpoints:\n", svc.Name)
+	if spec.Name == "" {
-	for _, endpoint := range svc.Endpoints() {
+		spec.Name, err = deploy.GenerateServiceName(spec.Container.Image)
-		fmt.Printf(" • %s\n", endpoint)
+		if err != nil {
+			return spec, fmt.Errorf("generate service name: %w", err)
+		}
 	}
 
-	return nil
+	return spec, err
+}
+
+// parseEnv parses the environment variables from the command line arguments.
+// It supports two formats: "VAR=value" or just "VAR" to use the value from the local environment.
+func parseEnv(env []string) (api.EnvVars, error) {
+	envVars := make(api.EnvVars)
+	for _, e := range env {
+		key, value, hasValue := strings.Cut(e, "=")
+		if key == "" {
+			return nil, fmt.Errorf("invalid environment variable: '%s'", e)
+		}
+
+		if hasValue {
+			envVars[key] = value
+		} else {
+			if localEnvValue, ok := os.LookupEnv(key); ok {
+				envVars[key] = localEnvValue
+			}
+		}
+	}
+
+	return envVars, nil
+}
+
+// parseVolumeFlags parses volume flag values in Docker CLI format and returns VolumeSpecs and VolumeMounts.
+// It handles both named volumes (volume_name:/container/path[:ro|volume-nocopy])
+// and bind mounts (/host/path:/container/path[:ro]).
+func parseVolumeFlags(volumes []string) ([]api.VolumeSpec, []api.VolumeMount, error) {
+	specs := make([]api.VolumeSpec, 0, len(volumes))
+	mounts := make([]api.VolumeMount, 0, len(volumes))
+
+	// Track volume names to avoid duplicate specs.
+	seenVolumes := make(map[string]struct{})
+
+	for _, vol := range volumes {
+		spec, mount, err := parseVolumeFlagValue(vol)
+		if err != nil {
+			return nil, nil, fmt.Errorf("invalid volume mount '%s': %w", vol, err)
+		}
+
+		if _, ok := seenVolumes[spec.Name]; !ok {
+			specs = append(specs, spec)
+			seenVolumes[spec.Name] = struct{}{}
+		}
+
+		mounts = append(mounts, mount)
+	}
+
+	return specs, mounts, nil
+}
+
+func parseVolumeFlagValue(volume string) (api.VolumeSpec, api.VolumeMount, error) {
+	var spec api.VolumeSpec
+	var mount api.VolumeMount
+
+	parts := strings.Split(volume, ":")
+	switch len(parts) {
+	case 1:
+		return spec, mount, fmt.Errorf("invalid format, must contain at least one separator ':'")
+	case 2, 3:
+		// Format: (volume_name|/host/path):/container/path[:opts]
+		if !strings.HasPrefix(parts[1], "/") {
+			return spec, mount, fmt.Errorf("invalid container mount path: '%s', must be absolute path", parts[1])
+		}
+
+		mount.ContainerPath = parts[1]
+		volumeNoCopy := false
+
+		if len(parts) == 3 {
+			opts := strings.Split(parts[2], ",")
+			for _, opt := range opts {
+				switch opt {
+				case "ro", "readonly":
+					mount.ReadOnly = true
+				case "volume-nocopy":
+					volumeNoCopy = true
+				default:
+					return spec, mount, fmt.Errorf("invalid option: '%s'", opt)
+				}
+			}
+		}
+
+		if strings.HasPrefix(parts[0], "/") {
+			// Host path bind mount: /host/path:/container/path
+			suffix, err := secret.RandomAlphaNumeric(4)
+			if err != nil {
+				return spec, mount, fmt.Errorf("generate random suffix: %w", err)
+			}
+
+			spec = api.VolumeSpec{
+				Name: "bind-" + suffix,
+				Type: api.VolumeTypeBind,
+				BindOptions: &api.BindOptions{
+					HostPath:       parts[0],
+					CreateHostPath: true,
+				},
+			}
+		} else {
+			// Named volume mount: volume_name:/container/path
+			volumeName := parts[0]
+			if !names.RestrictedNamePattern.MatchString(volumeName) {
+				return spec, mount, fmt.Errorf("volume name '%s' includes invalid characters, only '%s' are allowed. "+
+					"If you intended to pass a host directory or file, use absolute path",
+					volumeName, names.RestrictedNameChars)
+			}
+
+			spec = api.VolumeSpec{
+				Name: volumeName,
+				Type: api.VolumeTypeVolume,
+				VolumeOptions: &api.VolumeOptions{
+					Name:   volumeName,
+					NoCopy: volumeNoCopy,
+				},
+			}
+		}
+
+		mount.VolumeName = spec.Name
+	default:
+		return spec, mount, fmt.Errorf("invalid format, must container at most 2 separators ':'")
+	}
+
+	return spec, mount, nil
 }

cmd/uncloud/service/scale.go

@@ -0,0 +1,152 @@
+package service
+
+import (
+	"context"
+	"fmt"
+	"strconv"
+
+	"github.com/charmbracelet/huh"
+	"github.com/docker/compose/v2/pkg/progress"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/spf13/cobra"
+)
+
+type scaleOptions struct {
+	service  string
+	replicas uint
+	cluster  string
+}
+
+func NewScaleCommand() *cobra.Command {
+	opts := scaleOptions{}
+	cmd := &cobra.Command{
+		Use:   "scale SERVICE REPLICAS",
+		Short: "Scale a replicated service by changing the number of replicas.",
+		Long:  "Scale a replicated service by changing the number of replicas. Scaling down requires confirmation.",
+		Args:  cobra.ExactArgs(2),
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+
+			opts.service = args[0]
+			replicas, err := strconv.ParseUint(args[1], 10, 0)
+			if err != nil {
+				return fmt.Errorf("invalid number of replicas: %w", err)
+			}
+			opts.replicas = uint(replicas)
+
+			return scale(cmd.Context(), uncli, opts)
+		},
+	}
+
+	cmd.Flags().StringVarP(
+		&opts.cluster, "context", "c", "",
+		"Name of the cluster context. (default is the current context)",
+	)
+
+	return cmd
+}
+
+func scale(ctx context.Context, uncli *cli.CLI, opts scaleOptions) error {
+	if opts.replicas == 0 {
+		return fmt.Errorf(
+			"scaling to zero replicas is not supported. This would effectively remove the service without preserving "+
+				"its configuration, making it impossible to scale back up. Uncloud derives the service configuration "+
+				"from existing containers. Use 'uc rm %s' instead if you want to remove the service",
+			opts.service,
+		)
+	}
+
+	clusterClient, err := uncli.ConnectCluster(ctx, opts.cluster)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer clusterClient.Close()
+
+	svc, err := clusterClient.InspectService(ctx, opts.service)
+	if err != nil {
+		return fmt.Errorf("inspect service '%s': %w", opts.service, err)
+	}
+
+	if svc.Mode != api.ServiceModeReplicated {
+		return fmt.Errorf("scaling is only supported for services in %s mode, service '%s' is in %s mode",
+			api.ServiceModeReplicated, svc.Name, svc.Mode)
+	}
+
+	currentReplicas := uint(len(svc.Containers))
+
+	if currentReplicas == opts.replicas {
+		fmt.Printf("Service '%s' already has %d replicas. No changes required.\n", svc.Name, currentReplicas)
+		return nil
+	}
+
+	// TODO: Check if all containers have the same spec. If not, prompt user to choose which one to scale.
+	//  This can happen if a service deployment failed midway and some containers were not updated.
+	spec := svc.Containers[0].Container.ServiceSpec
+	spec.Replicas = opts.replicas
+	deployment := clusterClient.NewDeployment(spec, nil)
+	plan, err := deployment.Plan(ctx)
+	if err != nil {
+		return fmt.Errorf("plan deployment: %w", err)
+	}
+
+	if len(plan.Operations) == 0 {
+		fmt.Printf("Service '%s' is already scaled to %d replicas.\n", svc.Name, opts.replicas)
+		return nil
+	}
+
+	if opts.replicas < currentReplicas {
+		// Initialise a machine and container name resolver to properly format the plan output.
+		resolver, err := clusterClient.ServiceOperationNameResolver(ctx, svc)
+		if err != nil {
+			return fmt.Errorf("create machine and container name resolver for service operations: %w", err)
+		}
+
+		fmt.Printf("Scaling plan for service %s (%d → %d replicas):\n", svc.Name, currentReplicas, opts.replicas)
+		fmt.Println(plan.Format(resolver))
+		fmt.Println()
+
+		// Ask for confirmation before scaling down as it may cause data loss.
+		confirmed, err := confirm()
+		if err != nil {
+			return fmt.Errorf("confirm scaling: %w", err)
+		}
+		if !confirmed {
+			fmt.Println("Cancelled. No changes were made.")
+			return nil
+		}
+	}
+
+	title := fmt.Sprintf("Scaling service %s (%d → %d replicas)", svc.Name, currentReplicas, opts.replicas)
+	err = progress.RunWithTitle(ctx, func(ctx context.Context) error {
+		if _, err = deployment.Run(ctx); err != nil {
+			return fmt.Errorf("deploy service: %w", err)
+		}
+		return nil
+	}, uncli.ProgressOut(), title)
+	if err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func confirm() (bool, error) {
+	var confirmed bool
+	form := huh.NewForm(
+		huh.NewGroup(
+			huh.NewConfirm().
+				Title(
+					"Do you want to continue?",
+				).
+				Affirmative("Yes!").
+				Negative("No").
+				Value(&confirmed),
+		),
+	).WithAccessible(true)
+	if err := form.Run(); err != nil {
+		return false, err
+	}
+
+	return confirmed, nil
+}

cmd/uncloud/volume/create.go

@@ -0,0 +1,148 @@
+package volume
+
+import (
+	"context"
+	"fmt"
+	"slices"
+	"strings"
+
+	"github.com/charmbracelet/huh"
+	"github.com/docker/docker/api/types/volume"
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/spf13/cobra"
+)
+
+type createOptions struct {
+	driver     string
+	driverOpts []string
+	labels     []string
+	machine    string
+	context    string
+}
+
+func NewCreateCommand() *cobra.Command {
+	opts := createOptions{}
+
+	cmd := &cobra.Command{
+		Use:   "create VOLUME_NAME",
+		Short: "Create a volume on a specific machine.",
+		Args:  cobra.ExactArgs(1),
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+			opts.driver = strings.TrimSpace(opts.driver)
+
+			volumeName := args[0]
+			if volumeName == "" {
+				return fmt.Errorf("volume name is required")
+			}
+
+			return create(cmd.Context(), uncli, volumeName, opts)
+		},
+	}
+
+	cmd.Flags().StringVarP(&opts.driver, "driver", "d", "local",
+		"Volume driver to use.")
+	cmd.Flags().StringSliceVarP(&opts.driverOpts, "opt", "o", nil,
+		"Driver specific options in the form of 'key=value' pairs. Can be specified multiple times.")
+	cmd.Flags().StringSliceVarP(&opts.labels, "label", "l", nil,
+		"Labels to assign to the volume in the form of 'key=value' pairs. Can be specified multiple times.")
+	cmd.Flags().StringVarP(&opts.machine, "machine", "m", "",
+		"Name or ID of the machine to create the volume on.")
+	cmd.Flags().StringVarP(&opts.context, "context", "c", "",
+		"Name of the cluster context. (default is the current context)")
+
+	return cmd
+}
+
+func create(ctx context.Context, uncli *cli.CLI, name string, opts createOptions) error {
+	client, err := uncli.ConnectCluster(ctx, opts.context)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer client.Close()
+
+	// Parse driver options.
+	driverOpts := make(map[string]string)
+	for _, opt := range opts.driverOpts {
+		k, v, ok := strings.Cut(opt, "=")
+		if !ok {
+			return fmt.Errorf("invalid driver option format: '%s' (expected key=value)", opt)
+		}
+		driverOpts[k] = v
+	}
+
+	// Parse labels.
+	labels := make(map[string]string)
+	for _, label := range opts.labels {
+		k, v, ok := strings.Cut(label, "=")
+		if !ok {
+			return fmt.Errorf("invalid label format: '%s' (expected key=value)", label)
+		}
+		labels[k] = v
+	}
+
+	// List machines and filter by the specified machine name or ID.
+	// If no machine is specified, prompt the user to select one.
+	machines, err := client.ListMachines(ctx, nil)
+	if err != nil {
+		return fmt.Errorf("list machines: %w", err)
+	}
+	var selectedMachine *pb.MachineInfo
+
+	if opts.machine == "" {
+		if len(machines) == 1 {
+			selectedMachine = machines[0].Machine
+		} else {
+			if selectedMachine, err = promptSelectMachine(ctx, machines); err != nil {
+				return fmt.Errorf("select machine: %w", err)
+			}
+		}
+	} else {
+		m := machines.FindByNameOrID(opts.machine)
+		if m == nil {
+			return fmt.Errorf("machine '%s' not found", opts.machine)
+		}
+		selectedMachine = m.Machine
+	}
+
+	createOpts := volume.CreateOptions{
+		Name:       name,
+		Driver:     opts.driver,
+		DriverOpts: driverOpts,
+		Labels:     labels,
+	}
+	vol, err := client.CreateVolume(ctx, selectedMachine.Id, createOpts)
+	if err != nil {
+		return fmt.Errorf("create volume '%s' on machine '%s': %w", name, selectedMachine.Name, err)
+	}
+
+	fmt.Printf("Volume '%s' created on machine '%s'.\n", vol.Volume.Name, vol.MachineName)
+	return nil
+}
+
+func promptSelectMachine(ctx context.Context, machines api.MachineMembersList) (*pb.MachineInfo, error) {
+	options := make([]huh.Option[*pb.MachineInfo], len(machines))
+	for i, m := range machines {
+		options[i] = huh.NewOption(m.Machine.Name, m.Machine)
+	}
+	slices.SortFunc(options, func(a, b huh.Option[*pb.MachineInfo]) int {
+		return strings.Compare(a.Key, b.Key)
+	})
+
+	var selected *pb.MachineInfo
+	form := huh.NewForm(
+		huh.NewGroup(
+			huh.NewSelect[*pb.MachineInfo]().
+				Title("Select a machine to create the volume on (or specify with --machine flag)").
+				Options(options...).
+				Value(&selected),
+		),
+	)
+	if err := form.RunWithContext(ctx); err != nil {
+		return nil, err
+	}
+
+	return selected, nil
+}

cmd/uncloud/volume/inspect.go

@@ -0,0 +1,81 @@
+package volume
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/spf13/cobra"
+)
+
+type inspectOptions struct {
+	machine string
+	context string
+}
+
+func NewInspectCommand() *cobra.Command {
+	opts := inspectOptions{}
+
+	cmd := &cobra.Command{
+		Use:   "inspect VOLUME_NAME",
+		Short: "Display detailed information on a volume.",
+		Args:  cobra.ExactArgs(1),
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+			return inspect(cmd.Context(), uncli, args[0], opts)
+		},
+	}
+
+	cmd.Flags().StringVarP(&opts.machine, "machine", "m", "",
+		"Name or ID of the machine where the volume is located. "+
+			"If not specified, the volume will be searched across all machines.")
+	cmd.Flags().StringVarP(&opts.context, "context", "c", "",
+		"Name of the cluster context. (default is the current context)")
+
+	return cmd
+}
+
+func inspect(ctx context.Context, uncli *cli.CLI, name string, opts inspectOptions) error {
+	client, err := uncli.ConnectCluster(ctx, opts.context)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer client.Close()
+
+	filter := &api.VolumeFilter{
+		Names: []string{name},
+	}
+	if opts.machine != "" {
+		filter.Machines = []string{opts.machine}
+	}
+
+	volumes, err := client.ListVolumes(ctx, filter)
+	if err != nil {
+		return fmt.Errorf("list volumes: %w", err)
+	}
+
+	if len(volumes) == 0 {
+		if opts.machine != "" {
+			return fmt.Errorf("volume '%s' not found on machine '%s'", name, opts.machine)
+		}
+		return fmt.Errorf("volume '%s' not found on any machine", name)
+	}
+	if len(volumes) > 1 {
+		fmt.Printf("Volume '%s' found on multiple machines:\n", name)
+		for _, v := range volumes {
+			fmt.Printf(" • %s\n", v.MachineName)
+		}
+		return errors.New("specify --machine flag to choose which machine to use")
+	}
+
+	data, err := json.MarshalIndent(volumes[0], "", "  ")
+	if err != nil {
+		return fmt.Errorf("marshal volume: %w", err)
+	}
+	fmt.Println(string(data))
+
+	return nil
+}

cmd/uncloud/volume/ls.go

@@ -0,0 +1,105 @@
+package volume
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"slices"
+	"strings"
+	"text/tabwriter"
+
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/spf13/cobra"
+)
+
+type listOptions struct {
+	machines []string
+	quiet    bool
+	context  string
+}
+
+func NewListCommand() *cobra.Command {
+	opts := listOptions{}
+
+	cmd := &cobra.Command{
+		Use:     "ls",
+		Aliases: []string{"list"},
+		Short:   "List volumes across all machines in the cluster.",
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+			return list(cmd.Context(), uncli, opts)
+		},
+	}
+
+	cmd.Flags().StringSliceVarP(&opts.machines, "machine", "m", nil,
+		"Filter volumes by machine name or ID. Can be specified multiple times or as a comma-separated list. "+
+			"(default is include all machines)")
+	cmd.Flags().BoolVarP(&opts.quiet, "quiet", "q", false,
+		"Only display volume names.")
+
+	cmd.Flags().StringVarP(&opts.context, "context", "c", "",
+		"Name of the cluster context. (default is the current context)")
+
+	return cmd
+}
+
+func list(ctx context.Context, uncli *cli.CLI, opts listOptions) error {
+	client, err := uncli.ConnectCluster(ctx, opts.context)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer client.Close()
+
+	// Apply machine filter if specified.
+	var filter *api.VolumeFilter
+	if len(opts.machines) > 0 {
+		machines := cli.ExpandCommaSeparatedValues(opts.machines)
+		filter = &api.VolumeFilter{
+			Machines: machines,
+		}
+	}
+
+	volumes, err := client.ListVolumes(ctx, filter)
+	if err != nil {
+		return fmt.Errorf("list volumes: %w", err)
+	}
+
+	if len(volumes) == 0 {
+		if !opts.quiet {
+			fmt.Println("No volumes found.")
+		}
+		return nil
+	}
+
+	// Sort the volumes by name first, then by machine name.
+	slices.SortFunc(volumes, func(a, b api.MachineVolume) int {
+		cmp := strings.Compare(a.Volume.Name, b.Volume.Name)
+		if cmp != 0 {
+			return cmp
+		}
+		return strings.Compare(a.MachineName, b.MachineName)
+	})
+
+	// If quiet mode, just print volume names.
+	if opts.quiet {
+		for _, v := range volumes {
+			fmt.Println(v.Volume.Name)
+		}
+		return nil
+	}
+
+	// Print the volumes in a table format.
+	tw := tabwriter.NewWriter(os.Stdout, 0, 0, 3, ' ', 0)
+	fmt.Fprintln(tw, "NAME\tDRIVER\tMACHINE")
+
+	for _, v := range volumes {
+		fmt.Fprintf(tw, "%s\t%s\t%s\n",
+			v.Volume.Name,
+			v.Volume.Driver,
+			v.MachineName,
+		)
+	}
+
+	return tw.Flush()
+}

cmd/uncloud/volume/rm.go

@@ -0,0 +1,111 @@
+package volume
+
+import (
+	"context"
+	"errors"
+	"fmt"
+
+	"github.com/psviderski/uncloud/internal/cli"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/spf13/cobra"
+)
+
+type removeOptions struct {
+	force    bool
+	machines []string
+	yes      bool
+	context  string
+}
+
+func NewRemoveCommand() *cobra.Command {
+	opts := removeOptions{}
+
+	cmd := &cobra.Command{
+		Use:     "rm VOLUME_NAME [VOLUME_NAME...]",
+		Aliases: []string{"remove", "delete"},
+		Short:   "Remove one or more volumes.",
+		Long:    "Remove one or more volumes. You cannot remove a volume that is in use by a container.",
+		Args:    cobra.MinimumNArgs(1),
+		RunE: func(cmd *cobra.Command, args []string) error {
+			uncli := cmd.Context().Value("cli").(*cli.CLI)
+			return remove(cmd.Context(), uncli, args, opts)
+		},
+	}
+
+	cmd.Flags().BoolVarP(&opts.force, "force", "f", false,
+		"Force the removal of one or more volumes.")
+	cmd.Flags().StringSliceVarP(&opts.machines, "machine", "m", nil,
+		"Name or ID of the machine to remove one or more volumes from. "+
+			"Can be specified multiple times or as a comma-separated list.\n"+
+			"If not specified, the found volume(s) will be removed from all machines.")
+	cmd.Flags().BoolVarP(&opts.yes, "yes", "y", false,
+		"Do not prompt for confirmation before removing the volume(s).")
+
+	cmd.Flags().StringVarP(&opts.context, "context", "c", "",
+		"Name of the cluster context. (default is the current context)")
+
+	return cmd
+}
+
+func remove(ctx context.Context, uncli *cli.CLI, names []string, opts removeOptions) error {
+	client, err := uncli.ConnectCluster(ctx, opts.context)
+	if err != nil {
+		return fmt.Errorf("connect to cluster: %w", err)
+	}
+	defer client.Close()
+
+	filter := &api.VolumeFilter{
+		Names: names,
+	}
+
+	if len(opts.machines) > 0 {
+		machines := cli.ExpandCommaSeparatedValues(opts.machines)
+		filter.Machines = machines
+	}
+
+	volumes, err := client.ListVolumes(ctx, filter)
+	if err != nil {
+		return fmt.Errorf("list volumes: %w", err)
+	}
+
+	if len(volumes) == 0 {
+		if len(names) == 1 {
+			return fmt.Errorf("volume '%s' not found", names[0])
+		}
+		return fmt.Errorf("no volumes found matching the specified names")
+	}
+
+	// Confirm removal if not using --yes flag.
+	if !opts.yes {
+		fmt.Println("The following volumes will be removed:")
+		for _, v := range volumes {
+			fmt.Printf(" • '%s' on machine '%s'\n", v.Volume.Name, v.MachineName)
+		}
+
+		fmt.Println()
+		confirmed, err := cli.Confirm()
+		if err != nil {
+			return fmt.Errorf("confirm removal: %w", err)
+		}
+		if !confirmed {
+			fmt.Println("Cancelled. No volumes were removed.")
+			return nil
+		}
+	}
+
+	// Remove the volumes one by one collecting errors.
+	var removeErr error
+	for _, v := range volumes {
+		if err = client.RemoveVolume(ctx, v.MachineID, v.Volume.Name, opts.force); err != nil {
+			if !errors.Is(err, api.ErrNotFound) {
+				removeErr = errors.Join(removeErr, fmt.Errorf("failed to remove volume '%s' on machine '%s': %w",
+					v.Volume.Name, v.MachineName, err))
+			}
+			continue
+		}
+
+		fmt.Printf("Volume '%s' removed from machine '%s'.\n", v.Volume.Name, v.MachineName)
+	}
+
+	return removeErr
+}

cmd/uncloud/volume/root.go

@@ -0,0 +1,19 @@
+package volume
+
+import (
+	"github.com/spf13/cobra"
+)
+
+func NewRootCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "volume",
+		Short: "Manage volumes in an Uncloud cluster.",
+	}
+	cmd.AddCommand(
+		NewCreateCommand(),
+		NewInspectCommand(),
+		NewListCommand(),
+		NewRemoveCommand(),
+	)
+	return cmd
+}

cmd/uncloudd/main.go

@@ -2,14 +2,16 @@ package main
 
 import (
 	"context"
-	"github.com/spf13/cobra"
 	"log/slog"
 	"os"
 	"os/signal"
 	"syscall"
-	"uncloud/internal/daemon"
+
-	"uncloud/internal/log"
+	"github.com/psviderski/uncloud/internal/daemon"
-	"uncloud/internal/machine"
+	"github.com/psviderski/uncloud/internal/log"
+	"github.com/psviderski/uncloud/internal/machine"
+	"github.com/psviderski/uncloud/internal/version"
+	"github.com/spf13/cobra"
 )
 
 func main() {
@@ -22,6 +24,7 @@ func main() {
 	cmd := &cobra.Command{
 		Use:           "uncloudd",
 		Short:         "Uncloud machine daemon.",
+		Version:       version.String(),
 		SilenceUsage:  true,
 		SilenceErrors: true,
 		RunE: func(cmd *cobra.Command, args []string) error {
@@ -52,5 +55,4 @@ func main() {
 	}()
 
 	cobra.CheckErr(cmd.ExecuteContext(ctx))
-
 }

docs/.dockerignore

@@ -0,0 +1,14 @@
+# Ignore everything and use an allowlist of what is allowed to not package any secrets by accident.
+*
+
+# Allow files and directories.
+!docs/
+!src/
+!static/
+!pkg/
+!Caddyfile
+!*.js
+!*.json
+
+# Ignore unnecessary files inside allowed directories.
+**/.DS_Store

docs/.gitignore

@@ -0,0 +1,20 @@
+# Dependencies
+/node_modules
+
+# Production
+/build
+
+# Generated files
+.docusaurus
+.cache-loader
+
+# Misc
+.DS_Store
+.env.local
+.env.development.local
+.env.test.local
+.env.production.local
+
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*

docs/Caddyfile

@@ -0,0 +1,9 @@
+{
+    admin off
+}
+
+:8000 {
+    root * /usr/share/caddy
+    file_server
+    try_files {path} /index.html
+}

docs/Dockerfile

@@ -0,0 +1,27 @@
+# Stage 1: Base image.
+## Start with a base image containing NodeJS so we can build Docusaurus.
+FROM node:lts-alpine AS base
+## Disable colour output from yarn to make logs easier to read.
+ENV FORCE_COLOR=0
+## Enable corepack.
+RUN corepack enable
+## Set the working directory to `/opt/docusaurus`.
+WORKDIR /opt/docusaurus
+
+# Stage 2: Production build.
+FROM base AS prod
+## Set the working directory to `/opt/docusaurus`.
+WORKDIR /opt/docusaurus
+## Copy over the source code.
+COPY . /opt/docusaurus/
+## Install dependencies with `--immutable` to ensure reproducibility.
+RUN npm ci
+## Build the static site.
+RUN npm run build
+
+# Stage 3: Serve static site with Caddy.
+FROM caddy:2.10.0-alpine AS caddy
+## Copy the Caddyfile.
+COPY ./Caddyfile /etc/caddy/Caddyfile
+## Copy the Docusaurus build output.
+COPY --from=prod /opt/docusaurus/build /usr/share/caddy

docs/README.md

@@ -0,0 +1,41 @@
+# Website
+
+This website is built using [Docusaurus](https://docusaurus.io/), a modern static website generator.
+
+### Installation
+
+```
+$ yarn
+```
+
+### Local Development
+
+```
+$ yarn start
+```
+
+This command starts a local development server and opens up a browser window. Most changes are reflected live without having to restart the server.
+
+### Build
+
+```
+$ yarn build
+```
+
+This command generates static content into the `build` directory and can be served using any static contents hosting service.
+
+### Deployment
+
+Using SSH:
+
+```
+$ USE_SSH=true yarn deploy
+```
+
+Not using SSH:
+
+```
+$ GIT_USER=<Your GitHub username> yarn deploy
+```
+
+If you are using GitHub pages for hosting, this command is a convenient way to build the website and push to the `gh-pages` branch.

docs/compose.yaml

@@ -0,0 +1,8 @@
+services:
+  uncloud-docs:
+    image: ghcr.io/psviderski/uncloud-docs:latest
+    pull_policy: always
+    user: nobody
+    x-ports:
+      - docs.uncloud.run:8000/https
+    scale: 2

docs/docs/1-overview.md

@@ -0,0 +1,115 @@
+---
+slug: /
+---
+
+# Overview
+
+Uncloud makes **self-hosting web applications** across multiple machines in production dead simple.
+
+You can connect any machines — from cloud VMs to bare metal servers (no matter where they're located) — into a secure
+private network. Then run and scale multi-service and multi-container web apps and databases across your machines using
+simple Docker-like commands and [Docker Compose](https://docs.docker.com/reference/compose-file/) files.
+
+Uncloud covers all the essentials for operating apps in production without overwhelming you with the complexity of
+traditional container orchestrators like Kubernetes or Swarm:
+
+* Initial machine and network setup
+* Zero-downtime rolling deployments
+* Health checks and automatic restarts
+* Automatic HTTPS and reverse proxy configuration
+* Scaling services across multiple machines
+* Cross-machine service communication without exposing ports to the internet
+* DNS-based service discovery
+* Load balancing
+* Persistent storage
+
+## Use cases
+
+Some of the common use cases Uncloud is a great fit for:
+
+- **Self-hosting and Homelabs**: Run your self-hosted apps on your own hardware. Start with a single machine and add
+  more as your needs grow.
+- **Outgrowing Docker Compose**: Level up your Docker Compose setup with zero-downtime deployments, replicas across
+  multiple machines for improved reliability, cross-machine service communication, automated reverse proxy management,
+  and more using the same Compose file.
+- **Small to medium web applications**: Deploy your SaaS product, websites, or personal projects with redundancy across
+  multiple machines for better reliability and your peace of mind.
+- **Hybrid setups (cloud + on-prem)**: Combine cloud VMs with on-premise for cost savings and data sovereignty — all
+  managed through the same interface.
+- **Agencies and freelancers**: Host multiple client projects with proper isolation on shared infrastructure, optimising
+  costs and resources.
+- **Edge computing**: Deploy applications closer to your users for lower latency and better performance.
+- **Dev/staging environments**: Spin up additional environments for development and testing that mirror production
+  reusing the same Compose configuration.
+
+## What makes Uncloud different
+
+Here are the design decisions that make Uncloud stand out:
+
+### Decentralised design
+
+You can think of an Uncloud cluster as a **network of Docker hosts** (machines) that are all aware of each other. All
+machines in the cluster are equal. You can connect to any of them to manage containers on any other machine in the
+cluster. If a machine or part of the network goes down, the rest of the cluster keeps running.
+
+There is no centralised control plane, so no need to worry about maintaining a quorum of machines for it. The time saved
+can be better spent developing and deploying your apps instead.
+
+### Zero-config overlay network
+
+Uncloud automatically configures and maintains a secure **WireGuard mesh network** across your machines. It handles key
+management, peer discovery, and NAT traversal without any manual configuration. This makes it easy to connect machines
+from different networks and locations, such as cloud VMs, on-premise servers, or your Raspberry Pi at home.
+
+Docker containers running on different machines get **unique IP addresses** from the cluster network so they can
+**communicate directly** as if they were on a single machine without opening up any host ports to the internet.
+
+The design and implementation were highly inspired by
+Talos [KubeSpan](https://www.talos.dev/v1.10/talos-guides/network/kubespan/).
+
+### Managed DNS service
+
+Uncloud can provide **managed DNS records** like `<service-name>.<cluster-id>.cluster.uncloud.run` for your public
+services through free [Uncloud DNS](https://github.com/psviderski/uncloud-dns) service. You can deploy a service and
+instantly access it from anywhere with a proper DNS name and HTTPS without any manual DNS configuration. This makes
+self-hosting much more accessible and simplifies the process of adding your own domain later.
+
+### No complex orchestration
+
+Uncloud operations are done using **imperative CLI commands** that have the taste of Docker and Docker Compose. The
+deployment and scaling commands can output an execution plan that describes what exactly will be changed on your cluster
+once you approve it. For example, what containers and volumes will be created or removed, and on which machines.
+
+This gives you full visibility and control over every change with **immediate feedback** when something goes wrong.
+
+### Minimal resource footprint
+
+The Uncloud daemon consists of a couple Go and Rust binaries running alongside the Docker daemon on each machine. It
+needs no more than **150 MB of RAM** and a few percent of a CPU core in small setups. This minimal overhead maximises
+the system resources available for your apps.
+
+You can run Uncloud on machines with as little as 512 MB of RAM, assuming you also need some RAM for the OS and Docker,
+as well as the apps you want to run.
+
+### Troubleshooting-friendly
+
+When something goes wrong, you can dive straight into standard Docker containers without layers of abstraction in your
+way. You can also SSH into any machine and use the regular Linux troubleshooting tools. For example, `ping` service
+containers by their service names, `curl` service endpoints, or analyse traffic between containers using `wireshark`.
+
+## Getting started
+
+Install Uncloud CLI and deploy your first app in minutes:
+
+* [Install Uncloud CLI](./2-getting-started/1-install-cli.md)
+* [Deploy demo app](./2-getting-started/2-deploy-demo-app.md)
+
+## Getting help
+
+* **Discord community**: Join our [Discord server](https://discord.gg/eR35KQJhPu) for real-time discussions, support,
+  and updates.
+* **GitHub issues**: Report bugs or request features on our [GitHub repository](https://github.com/psviderski/uncloud).
+* **Documentation**: Browse the full documentation (this website you're on) for detailed guides and references. Use
+  search to find what you need quickly.
+* **Newsletter**: Subscribe to the [newsletter](https://uncloud.run/#subscribe) for development updates and early
+  insights.

docs/docs/2-getting-started/1-install-cli.md

@@ -0,0 +1,104 @@
+# Install CLI
+
+Install the Uncloud command-line utility to manage your machines and deploy apps using `uc` commands. You will run `uc`
+locally so choose the appropriate installation method for your operating system.
+
+:::info NOTE
+
+Windows is not natively supported yet, but you can install and run `uc` in a
+[WSL](https://learn.microsoft.com/en-us/windows/wsl/) terminal by following the instructions for Linux.
+:::
+
+## Homebrew (macOS, Linux)
+
+If you have [Homebrew](https://brew.sh/) package manager installed, this is the recommended installation method on macOS
+and Linux:
+
+```shell
+brew install psviderski/tap/uncloud
+```
+
+To upgrade to the latest version:
+
+```shell
+brew upgrade uncloud
+```
+
+## Install script (macOS, Linux)
+
+For a quick automated installation, use the install script:
+
+```shell
+curl -fsS https://get.uncloud.run/install.sh | sh
+```
+
+The script will:
+
+- Detect your operating system and architecture
+- Download the appropriate latest binary from [GitHub releases](https://github.com/psviderski/uncloud/releases)
+- Install it to `/usr/local/bin/uncloud` using `sudo` (you may need to enter your user password)
+- Create a shortcut `uc` in `/usr/local/bin` for convenience
+
+Don't like `curl | sh`? You can download and review the [install script](https://get.uncloud.run/install.sh) first and
+then run it:
+
+```shell
+curl -fsSO https://get.uncloud.run/install.sh
+cat install.sh
+sh install.sh
+```
+
+## GitHub download (macOS, Linux)
+
+You can manually download and use a pre-built binary from the
+[latest release](https://github.com/psviderski/uncloud/releases/latest) on GitHub.
+
+<Tabs>
+  <TabItem value="macOS (Apple Silicon)">
+    ```shell
+    curl -L https://github.com/psviderski/uncloud/releases/latest/download/uncloud_macos_arm64.tar.gz | tar xz
+    mv uncloud uc
+    ```
+  </TabItem>
+  <TabItem value="macOS (Intel)">
+    ```shell
+    curl -L https://github.com/psviderski/uncloud/releases/latest/download/uncloud_macos_amd64.tar.gz | tar xz
+    mv uncloud uc
+    ```
+  </TabItem>
+  <TabItem value="Linux (AMD 64-bit)">
+    ```shell
+    curl -L https://github.com/psviderski/uncloud/releases/latest/download/uncloud_linux_amd64.tar.gz | tar xz
+    mv uncloud uc
+    ```
+  </TabItem>
+  <TabItem value="Linux (ARM 64-bit)">
+    ```shell
+    curl -L https://github.com/psviderski/uncloud/releases/latest/download/uncloud_linux_arm64.tar.gz | tar xz
+    mv uncloud uc
+    ```
+  </TabItem>
+</Tabs>
+
+You can use the `./uc` binary directly from the current directory, or move it to a directory in your system's `PATH`
+to run it as `uc` from any location.
+
+For example, move it to `/usr/local/bin` which is a common location for user-installed binaries:
+
+```shell
+sudo mv ./uc /usr/local/bin
+```
+
+## Verify installation
+
+After installation, verify that `uc` command is working:
+
+```shell
+uc --version
+```
+
+## Next steps
+
+Now that you have `uc` installed, you're ready to:
+
+- [Deploy demo app](./2-deploy-demo-app.md)

docs/docs/2-getting-started/2-deploy-demo-app.md

@@ -0,0 +1,315 @@
+# Deploy demo app
+
+In this guide, we'll deploy [Excalidraw](https://excalidraw.com) — a popular sketching and diagramming tool — to your
+Linux server. You'll learn the **basics of Uncloud** and see how simple it is to **run web apps** on your own
+infrastructure with secure internet access.
+
+## Prerequisites
+
+Before you begin, you'll need:
+
+- **Uncloud CLI** [installed](1-install-cli.md) on your local machine
+- A **Ubuntu or Debian server** with **public IP address** and **SSH access** (as `root` or a user with `sudo`
+  privileges) using a **private key**.
+
+:::tip Need a server?
+
+A small Virtual Private Server (VPS) or dedicated server from providers like [Hetzner](https://www.hetzner.com) or
+[DigitalOcean](https://www.digitalocean.com) is a great choice for learning Uncloud and running lightweight services. We
+recommend using a freshly installed server as existing services on ports 80 and 443 can cause conflicts.
+
+**Minimum requirements:** 1 vCPU, 512 MB RAM, Ubuntu 22.04 or Debian 11, AMD64 (recommended) or ARM64 architecture.
+Other Linux distributions may work, but haven't been tested yet.
+
+:::
+
+## Set up your server
+
+First, let's turn your server into an Uncloud **machine**. This simply means setting it up so you can deploy and manage
+services on it using `uc`.
+
+```shell
+uc machine init root@<your-server-ip>
+```
+
+If the SSH key to access your server isn't added to your [SSH agent](https://www.ssh.com/academy/ssh/agent), specify it
+with the `-i` flag:
+
+```shell
+uc machine init root@<your-server-ip> -i ~/.ssh/id_xxx
+```
+
+This command will:
+
+- Install the latest stable Docker version on your server if it's not already installed
+- Install the Uncloud daemon on your server
+- Create a Docker network for Uncloud-managed containers
+- Deploy [Caddy](https://caddyserver.com/) as your reverse proxy listening on host ports 80 and 443
+- Reserve a free `xxxxxx.cluster.uncloud.run` subdomain via the Uncloud managed DNS service and point it to your
+  server's IP
+
+All in about a minute!
+
+<details>
+<summary>💡 Expand to see example output</summary>
+
+```
+$ uc machine init root@157.180.72.195
+Downloading Uncloud install script: https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/install.sh
+⏳ Running Uncloud install script...
+⏳ Installing Docker...
+# Executing docker install script, commit: 53a22f61c0628e58e1d6680b49e82993d304b449
++ sh -c apt-get -qq update >/dev/null
++ sh -c DEBIAN_FRONTEND=noninteractive apt-get -y -qq install ca-certificates curl >/dev/null
++ sh -c install -m 0755 -d /etc/apt/keyrings
++ sh -c curl -fsSL "https://download.docker.com/linux/ubuntu/gpg" -o /etc/apt/keyrings/docker.asc
++ sh -c chmod a+r /etc/apt/keyrings/docker.asc
++ sh -c echo "deb [arch=amd64 signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu noble stable" > /etc/apt/sources.list.d/docker.list
++ sh -c apt-get -qq update >/dev/null
++ sh -c DEBIAN_FRONTEND=noninteractive apt-get -y -qq install docker-ce docker-ce-cli containerd.io docker-compose-plugin docker-ce-rootless-extras docker-buildx-plugin >/dev/null
+
+Running kernel seems to be up-to-date.
+
+No services need to be restarted.
+
+No containers need to be restarted.
+
+No user sessions are running outdated binaries.
+
+No VM guests are running outdated hypervisor (qemu) binaries on this host.
++ sh -c docker version
+Client: Docker Engine - Community
+ Version:           28.2.1
+ API version:       1.50
+ Go version:        go1.24.3
+ Git commit:        879ac3f
+ Built:             Wed May 28 19:25:01 2025
+ OS/Arch:           linux/amd64
+ Context:           default
+
+Server: Docker Engine - Community
+ Engine:
+  Version:          28.2.1
+  API version:      1.50 (minimum version 1.24)
+  Go version:       go1.24.3
+  Git commit:       0e2cc22
+  Built:            Wed May 28 19:25:01 2025
+  OS/Arch:          linux/amd64
+  Experimental:     false
+ containerd:
+  Version:          1.7.27
+  GitCommit:        05044ec0a9a75232cad458027ca83437aae3f4da
+ runc:
+  Version:          1.2.5
+  GitCommit:        v1.2.5-0-g59923ef
+ docker-init:
+  Version:          0.19.0
+  GitCommit:        de40ad0
+
+================================================================================
+
+To run Docker as a non-privileged user, consider setting up the
+Docker daemon in rootless mode for your user:
+
+    dockerd-rootless-setuptool.sh install
+
+Visit https://docs.docker.com/go/rootless/ to learn about rootless mode.
+
+
+To run the Docker daemon as a fully privileged service, but granting non-root
+users access, refer to https://docs.docker.com/go/daemon-access/
+
+WARNING: Access to the remote API on a privileged Docker daemon is equivalent
+         to root access on the host. Refer to the 'Docker daemon attack surface'
+         documentation for details: https://docs.docker.com/go/attack-surface/
+
+================================================================================
+
+✓ Docker installed successfully.
+✓ Linux user and group 'uncloud' created.
+⏳ Installing Uncloud binaries...
+⏳ Downloading uncloudd binary: https://github.com/psviderski/uncloud/releases/latest/download/uncloudd_linux_amd64.tar.gz
+✓ uncloudd binary installed: /usr/local/bin/uncloudd
+⏳ Downloading uninstall script: https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/uninstall.sh
+✓ uncloud-uninstall script installed: /usr/local/bin/uncloud-uninstall
+✓ Systemd unit file created: /etc/systemd/system/uncloud.service
+Created symlink /etc/systemd/system/multi-user.target.wants/uncloud.service → /etc/systemd/system/uncloud.service.
+⏳ Downloading uncloud-corrosion binary: https://github.com/psviderski/corrosion/releases/latest/download/corrosion-x86_64-unknown-linux-gnu.tar.gz
+✓ uncloud-corrosion binary installed: /usr/local/bin/uncloud-corrosion
+✓ Systemd unit file created: /etc/systemd/system/uncloud-corrosion.service
+⏳ Starting Uncloud machine daemon (uncloud.service)...
+✓ Uncloud machine daemon started.
+✓ Uncloud installed on the machine successfully! 🎉
+Cluster initialised with machine 'machine-dc3c' and saved as context 'default' in your local config (/Users/spy/.config/uncloud/config.yaml)
+Current cluster context is now 'default'.
+Waiting for the machine to be ready...
+
+Reserved cluster domain: 7za6s7.cluster.uncloud.run
+[+] Deploying service caddy 7/2
+ ✔ Container caddy-d7uk on machine-dc3c     Started                                            6.1s
+   ✔ Image caddy:2.10.0 on machine-dc3c       Pulled                                           3.7s
+
+Updating cluster domain records in Uncloud DNS to point to machines running caddy service...
+[+] Verifying internet access to caddy service 1/1
+ ✔ Machine machine-dc3c (157.180.72.195)  Reachable                                            0.7s
+
+DNS records updated to use only the internet-reachable machines running caddy service:
+  *.7za6s7.cluster.uncloud.run  A → 157.180.72.195
+```
+
+</details>
+
+## Deploy Excalidraw
+
+Now that your machine is set up, let's deploy `excalidraw` service from the
+[official Docker image](https://hub.docker.com/r/excalidraw/excalidraw). The service will publish the container port 80
+as HTTPS endpoint on the previously reserved domain via Caddy.
+
+```shell
+uc run --name excalidraw --publish 80/https excalidraw/excalidraw
+```
+
+You'll see the progress of the deployment and the public URL where you can access the service:
+
+```
+[+] Running service excalidraw (replicated mode) 2/2
+ ✔ Container excalidraw-azpc on machine-dc3c  Started                       8.9s
+   ✔ Image excalidraw/excalidraw on machine-dc3c  Pulled                    4.7s
+
+excalidraw endpoints:
+ • https://excalidraw.7za6s7.cluster.uncloud.run → :80
+```
+
+## Verify your deployment
+
+After the service is deployed, use the `uc inspect` command to check its status and details:
+
+```shell
+uc inspect excalidraw
+```
+
+```
+ID:    4d2de1600b6ada221a03896cd388836c
+Name:  excalidraw
+Mode:  replicated
+
+CONTAINER ID   IMAGE                   CREATED              STATUS                        MACHINE
+fde7ac7f11ad   excalidraw/excalidraw   About a minute ago   Up About a minute (healthy)   machine-dc3c
+```
+
+In this example, the service has one container running on the machine `machine-dc3c` (our server). The container is up
+and healthy.
+
+You can also list all deployed services and their public endpoints using the `uc ls` command:
+
+```shell
+uc ls
+```
+
+```
+NAME         MODE         REPLICAS   ENDPOINTS
+caddy        global       1
+excalidraw   replicated   1          https://excalidraw.7za6s7.cluster.uncloud.run → :80
+```
+
+You can see `caddy` service listed here. That's your reverse proxy, running as a regular Uncloud service.
+
+## It's live! Start drawing! ✨
+
+Open your browser and navigate to the URL shown in the endpoints. It may take a moment for Caddy to obtain a TLS
+certificate from Let's Encrypt. If it doesn't load immediately, wait a few seconds and try again.
+
+![Excalidraw running on Uncloud](./img/excalidraw-browser.png)
+
+You now have:
+
+- Your **own Excalidraw instance** running on your server
+- A **public URL** with **automatic HTTPS** you can share with your team and friends
+- **Full control over your data** — no analytics or tracking
+
+## Convert to Docker Compose format
+
+Uncloud supports the [Compose file format](https://docs.docker.com/reference/compose-file/) for defining services. This
+allows you to version control your deployments, share configurations with your team, and deploy complex multi-service
+applications with a single command.
+
+Let's create a `compose.yaml` file in your current directory for the `excalidraw` service we just deployed.
+
+```yaml title="compose.yaml"
+services:
+  excalidraw:
+    image: excalidraw/excalidraw
+    x-ports:
+      - 80/https
+```
+
+:::info note
+
+The `x-ports` key is an Uncloud-specific extension to the Compose file format. It allows you to specify ports that
+should be published as HTTP(S) endpoints. Uncloud automatically configures the reverse proxy (Caddy) to route traffic to
+these ports.
+
+:::
+
+Now deploy it:
+
+```shell
+uc deploy
+```
+
+```
+Services are up to date.
+```
+
+Since `excalidraw` service is already running with the same configuration, Uncloud recognises there's nothing to change.
+We've successfully converted our deployment created with `uc run` to a Compose file.
+
+## Use your own domain
+
+Want to use your own domain, for example, `excalidraw.example.com` instead of `excalidraw.7za6s7.cluster.uncloud.run`?
+
+Add a CNAME record `excalidraw.example.com` in your DNS provider (Cloudflare, Namecheap, etc.) pointing to
+`excalidraw.7za6s7.cluster.uncloud.run`. Alternatively, you can add an A record pointing to your server's IP.
+
+Then update the published port `80/https` in `compose.yaml` to use your domain:
+
+```yaml title="compose.yaml"
+...
+x-ports:
+  - excalidraw.example.com:80/https
+```
+
+Finally, deploy the changes:
+
+```shell
+uc deploy
+```
+
+```
+Deployment plan:
+- Deploy service [name=excalidraw]
+  - machine-dc3c: Run container [image=excalidraw/excalidraw]
+  - machine-dc3c: Remove container [name=excalidraw-azpc]
+
+Do you want to continue?
+
+Choose [y/N]: y
+Chose: Yes!
+
+[+] Deploying services 2/2
+ ✔ Container excalidraw-0z12 on machine-dc3c  Started                       3.5s 
+ ✔ Container excalidraw-azpc on machine-dc3c  Removed                       3.4s 
+```
+
+Notice how Uncloud performed a **zero-downtime deployment** — it started the new container with the updated
+configuration before removing the old one. Your service stayed available throughout the update.
+
+Give it a moment for Caddy to obtain a TLS certificate, then visit https://excalidraw.example.com.
+
+## Clean up
+
+TBD
+
+## Next steps
+
+TBD

docs/docs/2-getting-started/_category_.yaml

@@ -0,0 +1,4 @@
+label: Getting started
+collapsed: false # keep the category open by default
+link:
+  type: generated-index

docs/docusaurus.config.js

@@ -0,0 +1,153 @@
+// @ts-check
+// `@type` JSDoc annotations allow editor autocompletion and type checking
+// (when paired with `@ts-check`).
+// There are various equivalent ways to declare your Docusaurus config.
+// See: https://docusaurus.io/docs/api/docusaurus-config
+
+import {themes as prismThemes} from 'prism-react-renderer';
+
+// This runs in Node.js - Don't use client-side code here (browser APIs, JSX...)
+
+/** @type {import('@docusaurus/types').Config} */
+const config = {
+  title: 'Uncloud Docs',
+  tagline: 'Dinosaurs are cool',
+  favicon: 'img/favicon.png',
+
+  // Set the production url of your site here
+  url: 'https://docs.uncloud.run',
+  // Set the /<baseUrl>/ pathname under which your site is served
+  // For GitHub pages deployment, it is often '/<projectName>/'
+  baseUrl: '/',
+
+  // GitHub pages deployment config.
+  // If you aren't using GitHub pages, you don't need these.
+  organizationName: '', // Usually your GitHub org/user name.
+  projectName: '', // Usually your repo name.
+
+  onBrokenLinks: 'throw',
+  onBrokenMarkdownLinks: 'warn',
+
+  // Even if you don't use internationalization, you can use this field to set
+  // useful metadata like html lang. For example, if your site is Chinese, you
+  // may want to replace "en" with "zh-Hans".
+  i18n: {
+    defaultLocale: 'en',
+    locales: ['en'],
+  },
+
+  themes: [
+    [
+      '@easyops-cn/docusaurus-search-local',
+      /** @type {import("@easyops-cn/docusaurus-search-local").PluginOptions} */
+      ({
+        docsRouteBasePath: '/',
+        hashed: true,
+        highlightSearchTermsOnTargetPage: true,
+      }),
+    ],
+  ],
+  presets: [
+    [
+      'classic',
+      /** @type {import('@docusaurus/preset-classic').Options} */
+      ({
+        docs: {
+          // Remove this to remove the "edit this page" links.
+          editUrl: 'https://github.com/psviderski/uncloud/edit/main/docs/',
+          // Serve the docs at the site's root.
+          routeBasePath: '/',
+          showLastUpdateTime: true,
+          sidebarPath: './sidebars.js',
+        },
+        // blog: {
+        //   showReadingTime: true,
+        //   feedOptions: {
+        //     type: ['rss', 'atom'],
+        //     xslt: true,
+        //   },
+        //   // Please change this to your repo.
+        //   // Remove this to remove the "edit this page" links.
+        //   editUrl:
+        //     'https://github.com/facebook/docusaurus/tree/main/packages/create-docusaurus/templates/shared/',
+        //   // Useful options to enforce blogging best practices
+        //   onInlineTags: 'warn',
+        //   onInlineAuthors: 'warn',
+        //   onUntruncatedBlogPosts: 'warn',
+        // },
+        theme: {
+          customCss: './src/css/custom.css',
+        },
+      }),
+    ],
+  ],
+
+  themeConfig:
+    /** @type {import('@docusaurus/preset-classic').ThemeConfig} */
+    ({
+      colorMode: {
+        // The color mode when user first visits the site.
+        defaultMode: "light",
+        // Whether to use the prefers-color-scheme media-query, using user system preferences,
+        // instead of the hardcoded defaultMode.
+        respectPrefersColorScheme: true,
+      },
+      // The meta image URL for the site (og:image and twitter:image meta tags).
+      // Relative to your site's "static" directory. Cannot be SVGs. Can be external URLs too.
+      image: 'img/social-card.png',
+      navbar: {
+        title: 'Uncloud Docs',
+        logo: {
+          alt: 'Uncloud Logo',
+          src: 'img/logo.svg',
+        },
+        items: [
+          {
+            type: 'search',
+            position: 'right',
+          },
+          {
+            href: 'https://github.com/psviderski/uncloud',
+            label: 'GitHub',
+            position: 'right',
+          },
+        ],
+      },
+      footer: {
+        style: 'light',
+        links: [
+          {
+            title: 'Uncloud',
+            items: [
+              {
+                html: 'Made with ❤️ by <a href="https://github.com/psviderski">@psviderski</a> in Australia'
+              },
+            ],
+          },
+          {
+            title: 'Community',
+            items: [
+              {
+                label: 'Discord',
+                href: 'https://discord.gg/eR35KQJhPu',
+              },
+              {
+                label: 'X',
+                href: 'https://x.com/psviderski',
+              },
+              {
+                label: 'GitHub',
+                href: 'https://github.com/psviderski/uncloud',
+              },
+            ],
+          }
+        ]
+      },
+      prism: {
+        theme: prismThemes.palenight,
+        //darkTheme: prismThemes.dracula,
+      },
+    }),
+};
+
+export default config;

docs/package.json

@@ -0,0 +1,45 @@
+{
+  "name": "docs",
+  "version": "0.0.0",
+  "private": true,
+  "scripts": {
+    "docusaurus": "docusaurus",
+    "start": "docusaurus start",
+    "build": "docusaurus build",
+    "swizzle": "docusaurus swizzle",
+    "deploy": "docusaurus deploy",
+    "clear": "docusaurus clear",
+    "serve": "docusaurus serve",
+    "write-translations": "docusaurus write-translations",
+    "write-heading-ids": "docusaurus write-heading-ids"
+  },
+  "dependencies": {
+    "@docusaurus/core": "3.7.0",
+    "@docusaurus/preset-classic": "3.7.0",
+    "@easyops-cn/docusaurus-search-local": "^0.49.2",
+    "@mdx-js/react": "^3.0.0",
+    "clsx": "^2.0.0",
+    "prism-react-renderer": "^2.3.0",
+    "react": "^19.0.0",
+    "react-dom": "^19.0.0"
+  },
+  "devDependencies": {
+    "@docusaurus/module-type-aliases": "3.7.0",
+    "@docusaurus/types": "3.7.0"
+  },
+  "browserslist": {
+    "production": [
+      ">0.5%",
+      "not dead",
+      "not op_mini all"
+    ],
+    "development": [
+      "last 3 chrome version",
+      "last 3 firefox version",
+      "last 5 safari version"
+    ]
+  },
+  "engines": {
+    "node": ">=18.0"
+  }
+}

docs/sidebars.js

@@ -0,0 +1,35 @@
+// @ts-check
+
+// This runs in Node.js - Don't use client-side code here (browser APIs, JSX...)
+
+/**
+ * Creating a sidebar enables you to:
+ - create an ordered group of docs
+ - render a sidebar for each doc of that group
+ - provide next/previous navigation
+
+ The sidebars can be generated from the filesystem, or explicitly defined here.
+
+ Create as many sidebars as you want.
+
+ @type {import('@docusaurus/plugin-content-docs').SidebarsConfig}
+ */
+const sidebars = {
+  // By default, Docusaurus generates a sidebar from the docs folder structure
+  sidebar: [{type: 'autogenerated', dirName: '.'}],
+
+  // But you can create a sidebar manually
+  /*
+  tutorialSidebar: [
+    'intro',
+    'hello',
+    {
+      type: 'category',
+      label: 'Tutorial',
+      items: ['tutorial-basics/create-a-document'],
+    },
+  ],
+   */
+};
+
+export default sidebars;

docs/src/css/custom.css

@@ -0,0 +1,40 @@
+/**
+ * Any CSS included here will be global. The classic template
+ * bundles Infima by default. Infima is a CSS framework designed to
+ * work well for content-centric websites.
+ */
+
+@import url('https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;700&display=swap');
+
+/* You can override the default Infima variables here. */
+:root {
+  --ifm-font-family-base: 'Inter', sans-serif;
+  --ifm-heading-font-family: 'Inter', sans-serif;
+
+  --ifm-color-primary: #006be6;
+  --ifm-color-primary-dark: #0060cf;
+  --ifm-color-primary-darker: #005bc4;
+  --ifm-color-primary-darkest: #004ba1;
+  --ifm-color-primary-light: #0076fd;
+  --ifm-color-primary-lighter: #0a7cff;
+  --ifm-color-primary-lightest: #2c8eff;
+  --ifm-code-font-size: 95%;
+  --docusaurus-highlighted-code-line-bg: rgba(0, 0, 0, 0.1);
+}
+
+/* For readability concerns, you should choose a lighter palette in dark mode. */
+[data-theme='dark'] {
+  --ifm-color-primary: #3391ee;
+  --ifm-color-primary-dark: #1883ec;
+  --ifm-color-primary-darker: #137be3;
+  --ifm-color-primary-darkest: #1066bb;
+  --ifm-color-primary-light: #4e9ff0;
+  --ifm-color-primary-lighter: #5ba7f1;
+  --ifm-color-primary-lightest: #83bcf5;
+  --docusaurus-highlighted-code-line-bg: rgba(0, 0, 0, 0.3);
+}
+
+.menu {
+  font-size: 0.875rem;
+  font-weight: var(--ifm-font-weight-normal);
+}

docs/src/theme/MDXComponents.js

@@ -0,0 +1,13 @@
+import React from 'react';
+// Import the original mapper
+import MDXComponents from '@theme-original/MDXComponents';
+import Tabs from '@theme/Tabs';
+import TabItem from '@theme/TabItem';
+
+export default {
+    // Re-use the default mapping
+    ...MDXComponents,
+    // Global import the <Tabs> and <TabItem> components
+    Tabs,
+    TabItem,
+};

docs/static/img/logo.svg

@@ -0,0 +1,7 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 80 80">
+    <g>
+        <rect x="0" y="0" width="80" height="80" rx="10" fill="#18181B"/>
+        <path d="M20 32 L20 48 L36 48 L36 32 M60 32 L45 32 L45 48 L60 48" stroke="white" stroke-width="6" fill="none"
+              stroke-linecap="square" stroke-linejoin="miter"/>
+    </g>
+</svg>

docs/user_guide.md

@@ -0,0 +1,59 @@
+# User Guide
+
+## Initialising a cluster
+
+To begin setting up a new Uncloud cluster, create the desired nodes. Ensure that their firewall allows the required ports.
+
+### Ports
+
+Nodes running Uncloud with a standard configuration must have the following inbound ports allowed in the firewall:
+
+* 51820/udp (WireGuard meshnet)
+* 22/tcp (default SSH management port)
+* 80/tcp (required for challenge if uncloud.run DNS is enabled, pass --no-dns during init to disable)
+
+In addition, any ports for any workloads you want to expose will need to be open.
+
+### Configuration
+
+Uncloud stores its configuration in `~/.config/uncloud/config.yaml`. If you wish to reinitialise a cluster, simply remove it from this config.
+
+### Initialisation
+
+Begin by initialising the first node in your cluster with `uc machine init [USER@HOST:PORT]`. If you do not have a need for Caddy reverse proxy, you may disable this feature with `--no-caddy`.
+
+This command will idempotently install Docker, uncloudd, uncloud-corrosion. If Caddy is enabled, it will set up a reverse proxy. If Uncloud DNS is enabled, it will create a DNS A record for the machine's public IP address under `*.[CLUSTER ID].cluster.uncloud.run`.
+
+If you wish to uninstall Uncloud and its components, run `uncloud-uninstall`.
+
+### Adding a node
+
+Just like the initialisation of the first node, a node can be added to the cluster with `uc machine add`.
+
+### DNS
+
+Uncloud (uncloud.run) DNS can be managed with the `uc dns` subcommand.
+
+* To reserve a domain name, run `uc dns reserve`
+* To release a domain name, run `uc dns release`.
+* To see the domain name, run `uc dns show`
+
+### Running a service
+
+Services on an Uncloud cluster can be managed with `uc service`.
+
+You can run a service with two replicas that expose port 80 like the following:
+
+```
+uc service run -p 80/http --replicas 2 nginxdemos/hello
+```
+
+This requires the Caddy reverse proxy to be deployed. If it wasn't previously, it can be deployed with `uc caddy deploy`.
+
+Since this doesn't specify a service name, a random one will be generated.
+
+The service can be deleted by its service name:
+
+```
+uc service rm hello-gsdo
+```

experiment/talos_discovery.go

@@ -11,7 +11,7 @@ import (
 	"go.uber.org/zap"
 	"net/netip"
 	"time"
-	"uncloud/internal/machine/network"
+	"github.com/psviderski/uncloud/internal/machine/network"
 )
 
 const (

go.mod

@@ -1,4 +1,4 @@
-module uncloud
+module github.com/psviderski/uncloud
 
 go 1.23.0
 
@@ -11,7 +11,10 @@ require (
 	github.com/caddyserver/caddy/v2 v2.8.4
 	github.com/cenkalti/backoff/v4 v4.3.0
 	github.com/charmbracelet/huh v0.6.0
+	github.com/charmbracelet/lipgloss v0.13.0
+	github.com/compose-spec/compose-go/v2 v2.4.5
 	github.com/coreos/go-systemd v0.0.0-20191104093116-d3cd4ed1dbcf
+	github.com/deckarep/golang-set/v2 v2.8.0
 	github.com/dgraph-io/badger/v3 v3.2103.5
 	github.com/distribution/reference v0.6.0
 	github.com/docker/cli v27.5.0+incompatible
@@ -19,6 +22,8 @@ require (
 	github.com/docker/docker v27.4.0-rc.2+incompatible
 	github.com/docker/go-connections v0.5.0
 	github.com/docker/go-units v0.5.0
+	github.com/goccy/go-yaml v1.17.1
+	github.com/google/go-cmp v0.7.0
 	github.com/google/go-containerregistry v0.20.2
 	github.com/hashicorp/memberlist v0.5.1
 	github.com/hashicorp/serf v0.10.1
@@ -29,7 +34,11 @@ require (
 	github.com/ipfs/go-ds-crdt v0.5.2
 	github.com/ipfs/go-ipld-format v0.6.0
 	github.com/ipfs/go-log/v2 v2.5.1
+	github.com/jmoiron/sqlx v1.4.0
 	github.com/lmittmann/tint v1.0.5
+	github.com/miekg/dns v1.1.65
+	github.com/moby/term v0.5.0
+	github.com/opencontainers/go-digest v1.0.0
 	github.com/opencontainers/image-spec v1.1.0
 	github.com/siderolabs/discovery-api v0.1.4
 	github.com/siderolabs/discovery-client v0.1.9
@@ -39,15 +48,16 @@ require (
 	github.com/vishvananda/netlink v1.3.0
 	go.uber.org/zap v1.27.0
 	go4.org/netipx v0.0.0-20231129151722-fdeea329fbba
-	golang.org/x/crypto v0.32.0
+	golang.org/x/crypto v0.33.0
-	golang.org/x/net v0.34.0
+	golang.org/x/net v0.35.0
-	golang.org/x/sync v0.10.0
+	golang.org/x/sync v0.11.0
-	golang.org/x/sys v0.29.0
+	golang.org/x/sys v0.31.0
 	golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173
 	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6
 	google.golang.org/genproto/googleapis/rpc v0.0.0-20241209162323-e6fa225c2576
 	google.golang.org/grpc v1.68.1
 	google.golang.org/protobuf v1.36.3
+	modernc.org/sqlite v1.36.3
 )
 
 require (
@@ -77,13 +87,11 @@ require (
 	github.com/cespare/xxhash/v2 v2.3.0 // indirect
 	github.com/charmbracelet/bubbles v0.20.0 // indirect
 	github.com/charmbracelet/bubbletea v1.1.1 // indirect
-	github.com/charmbracelet/lipgloss v0.13.0 // indirect
 	github.com/charmbracelet/x/ansi v0.3.2 // indirect
 	github.com/charmbracelet/x/exp/strings v0.0.0-20240919170804-a4978c8e603a // indirect
 	github.com/charmbracelet/x/term v0.2.0 // indirect
 	github.com/chzyer/readline v1.5.1 // indirect
 	github.com/cloudflare/cfssl v1.6.4 // indirect
-	github.com/compose-spec/compose-go/v2 v2.4.5 // indirect
 	github.com/containerd/console v1.0.4 // indirect
 	github.com/containerd/containerd v1.7.24 // indirect
 	github.com/containerd/containerd/api v1.7.19 // indirect
@@ -118,7 +126,7 @@ require (
 	github.com/go-logfmt/logfmt v0.6.0 // indirect
 	github.com/go-logr/logr v1.4.2 // indirect
 	github.com/go-logr/stdr v1.2.2 // indirect
-	github.com/go-sql-driver/mysql v1.7.1 // indirect
+	github.com/go-sql-driver/mysql v1.8.1 // indirect
 	github.com/go-task/slim-sprig/v3 v3.0.0 // indirect
 	github.com/go-viper/mapstructure/v2 v2.0.0 // indirect
 	github.com/godbus/dbus/v5 v5.1.0 // indirect
@@ -131,8 +139,7 @@ require (
 	github.com/google/btree v1.1.2 // indirect
 	github.com/google/cel-go v0.20.1 // indirect
 	github.com/google/flatbuffers v24.3.25+incompatible // indirect
-	github.com/google/go-cmp v0.6.0 // indirect
+	github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e // indirect
-	github.com/google/pprof v0.0.0-20241210010833-40e02aabc2ad // indirect
 	github.com/google/shlex v0.0.0-20191202100458-e7afc7fbc510 // indirect
 	github.com/google/uuid v1.6.0 // indirect
 	github.com/gorilla/mux v1.8.1 // indirect
@@ -192,7 +199,6 @@ require (
 	github.com/mdlayher/socket v0.5.1 // indirect
 	github.com/mgutz/ansi v0.0.0-20200706080929-d51e80ef957d // indirect
 	github.com/mholt/acmez/v2 v2.0.3 // indirect
-	github.com/miekg/dns v1.1.62 // indirect
 	github.com/miekg/pkcs11 v1.1.1 // indirect
 	github.com/minio/sha256-simd v1.0.1 // indirect
 	github.com/mitchellh/cli v1.1.5 // indirect
@@ -208,7 +214,8 @@ require (
 	github.com/moby/patternmatcher v0.6.0 // indirect
 	github.com/moby/sys/sequential v0.6.0 // indirect
 	github.com/moby/sys/signal v0.7.1 // indirect
-	github.com/moby/term v0.5.0 // indirect
+	github.com/moby/sys/user v0.3.0 // indirect
+	github.com/moby/sys/userns v0.1.0 // indirect
 	github.com/morikuni/aec v1.0.0 // indirect
 	github.com/mr-tron/base58 v1.2.0 // indirect
 	github.com/muesli/ansi v0.0.0-20230316100256-276c6243b2f6 // indirect
@@ -224,8 +231,8 @@ require (
 	github.com/multiformats/go-multistream v0.5.0 // indirect
 	github.com/multiformats/go-varint v0.0.7 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
+	github.com/ncruces/go-strftime v0.1.9 // indirect
 	github.com/onsi/ginkgo/v2 v2.22.0 // indirect
-	github.com/opencontainers/go-digest v1.0.0 // indirect
 	github.com/pelletier/go-toml v1.9.5 // indirect
 	github.com/pires/go-proxyproto v0.7.0 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
@@ -239,6 +246,7 @@ require (
 	github.com/prometheus/procfs v0.15.1 // indirect
 	github.com/quic-go/qpack v0.5.1 // indirect
 	github.com/quic-go/quic-go v0.48.2 // indirect
+	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
 	github.com/rivo/uniseg v0.4.7 // indirect
 	github.com/rs/xid v1.5.0 // indirect
 	github.com/russross/blackfriday/v2 v2.1.0 // indirect
@@ -269,6 +277,9 @@ require (
 	github.com/urfave/cli v1.22.16 // indirect
 	github.com/vbatts/tar-split v0.11.6 // indirect
 	github.com/vishvananda/netns v0.0.4 // indirect
+	github.com/xeipuuv/gojsonpointer v0.0.0-20190905194746-02993c407bfb // indirect
+	github.com/xeipuuv/gojsonreference v0.0.0-20180127040603-bd5ef7bd5415 // indirect
+	github.com/xeipuuv/gojsonschema v1.2.0 // indirect
 	github.com/zeebo/blake3 v0.2.4 // indirect
 	go.etcd.io/bbolt v1.3.11 // indirect
 	go.opencensus.io v0.24.0 // indirect
@@ -294,11 +305,11 @@ require (
 	go.uber.org/zap/exp v0.2.0 // indirect
 	golang.org/x/crypto/x509roots/fallback v0.0.0-20240507223354-67b13616a595 // indirect
 	golang.org/x/exp v0.0.0-20241215155358-4a5509556b9e // indirect
-	golang.org/x/mod v0.22.0 // indirect
+	golang.org/x/mod v0.23.0 // indirect
-	golang.org/x/term v0.28.0 // indirect
+	golang.org/x/term v0.29.0 // indirect
-	golang.org/x/text v0.21.0 // indirect
+	golang.org/x/text v0.22.0 // indirect
 	golang.org/x/time v0.8.0 // indirect
-	golang.org/x/tools v0.29.0 // indirect
+	golang.org/x/tools v0.30.0 // indirect
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect
 	google.golang.org/genproto v0.0.0-20240401170217-c3f982113cda // indirect
 	google.golang.org/genproto/googleapis/api v0.0.0-20241209162323-e6fa225c2576 // indirect
@@ -306,4 +317,7 @@ require (
 	gvisor.dev/gvisor v0.0.0-20230927004350-cbd86285d259 // indirect
 	howett.net/plist v1.0.0 // indirect
 	lukechampine.com/blake3 v1.3.0 // indirect
+	modernc.org/libc v1.61.13 // indirect
+	modernc.org/mathutil v1.7.1 // indirect
+	modernc.org/memory v1.8.2 // indirect
 )

go.sum

@@ -233,6 +233,8 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davidlazar/go-crypto v0.0.0-20200604182044-b73af7476f6c h1:pFUpOrbxDR6AkioZ1ySsx5yxlDQZ8stG2b88gTPxgJU=
 github.com/davidlazar/go-crypto v0.0.0-20200604182044-b73af7476f6c/go.mod h1:6UhI8N9EjYm1c2odKpFpAYeR8dsBeM7PtzQhRgxRr9U=
+github.com/deckarep/golang-set/v2 v2.8.0 h1:swm0rlPCmdWn9mESxKOjWk8hXSqoxOp+ZlfuyaAdFlQ=
+github.com/deckarep/golang-set/v2 v2.8.0/go.mod h1:VAky9rY/yGXJOLEDv3OMci+7wtDpOF4IN+y82NBOac4=
 github.com/decred/dcrd/crypto/blake256 v1.0.1 h1:7PltbUIQB7u/FfZ39+DGa/ShuMyJ5ilcvdfma9wOH6Y=
 github.com/decred/dcrd/crypto/blake256 v1.0.1/go.mod h1:2OfgNZ5wDpcsFmHmCK5gZTPcCXqlm2ArzUIkw9czNJo=
 github.com/decred/dcrd/dcrec/secp256k1/v4 v4.3.0 h1:rpfIENRNNilwHwZeG5+P150SMrnNEcHYvcCuK6dPZSg=
@@ -332,8 +334,8 @@ github.com/go-logr/logr v1.4.2/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ4
 github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
 github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
 github.com/go-sql-driver/mysql v1.3.0/go.mod h1:zAC/RDZ24gD3HViQzih4MyKcchzm+sOG5ZlKdlhCg5w=
-github.com/go-sql-driver/mysql v1.7.1 h1:lUIinVbN1DY0xBg0eMOzmmtGoHwWBbvnWubQUrtU8EI=
+github.com/go-sql-driver/mysql v1.8.1 h1:LedoTUt/eveggdHS9qUFC1EFSa8bU2+1pZjSRpvNJ1Y=
-github.com/go-sql-driver/mysql v1.7.1/go.mod h1:OXbVy3sEdcQ2Doequ6Z5BW6fXNQTmx+9S1MCJN5yJMI=
+github.com/go-sql-driver/mysql v1.8.1/go.mod h1:wEBSXgmK//2ZFJyE+qWnIsVGmvmEKlqwuVSjsCm7DZg=
 github.com/go-stack/stack v1.6.0/go.mod h1:v0f6uXyyMGvRgIKkXu+yp6POWl0qKG85gN/melR3HDY=
 github.com/go-stack/stack v1.8.0/go.mod h1:v0f6uXyyMGvRgIKkXu+yp6POWl0qKG85gN/melR3HDY=
 github.com/go-task/slim-sprig/v3 v3.0.0 h1:sUs3vkvUymDpBKi3qH1YSqBQk9+9D/8M2mN1vB6EwHI=
@@ -341,6 +343,8 @@ github.com/go-task/slim-sprig/v3 v3.0.0/go.mod h1:W848ghGpv3Qj3dhTPRyJypKRiqCdHZ
 github.com/go-viper/mapstructure/v2 v2.0.0 h1:dhn8MZ1gZ0mzeodTG3jt5Vj/o87xZKuNAprG2mQfMfc=
 github.com/go-viper/mapstructure/v2 v2.0.0/go.mod h1:oJDH3BJKyqBA2TXFhDsKDGDTlndYOZ6rGS0BRZIxGhM=
 github.com/go-yaml/yaml v2.1.0+incompatible/go.mod h1:w2MrLa16VYP0jy6N7M5kHaCkaLENm+P+Tv+MfurjSw0=
+github.com/goccy/go-yaml v1.17.1 h1:LI34wktB2xEE3ONG/2Ar54+/HJVBriAGJ55PHls4YuY=
+github.com/goccy/go-yaml v1.17.1/go.mod h1:XBurs7gK8ATbW4ZPGKgcbrY1Br56PdM69F7LkFRi1kA=
 github.com/godbus/dbus/v5 v5.1.0 h1:4KLkAxT3aOY8Li4FRJe/KvhoNFFxo0m6fNuFUO8QJUk=
 github.com/godbus/dbus/v5 v5.1.0/go.mod h1:xhWf0FNVPg57R7Z0UbKHbJfkEywrmjJnf7w5xrFpKfA=
 github.com/gofrs/flock v0.12.1 h1:MTLVXXHf8ekldpJk3AKicLij9MdwOWkZ+a/jHHZby9E=
@@ -395,8 +399,8 @@ github.com/google/go-cmp v0.5.0/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/
 github.com/google/go-cmp v0.5.3/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.4/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.9/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
-github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
-github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
 github.com/google/go-containerregistry v0.20.2 h1:B1wPJ1SN/S7pB+ZAimcciVD+r+yV/l/DSArMxlbwseo=
 github.com/google/go-containerregistry v0.20.2/go.mod h1:z38EKdKh4h7IP2gSfUUqEvalZBqs6AoLeWfUy34nQC8=
 github.com/google/go-tpm v0.9.0 h1:sQF6YqWMi+SCXpsmS3fd21oPy/vSddwZry4JnmltHVk=
@@ -408,8 +412,8 @@ github.com/google/go-tspi v0.3.0/go.mod h1:xfMGI3G0PhxCdNVcYr1C4C+EizojDg/TXuX5b
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
 github.com/google/gopacket v1.1.19 h1:ves8RnFZPGiFnTS0uPQStjwru6uO6h+nlr9j6fL7kF8=
 github.com/google/gopacket v1.1.19/go.mod h1:iJ8V8n6KS+z2U1A8pUwu8bW5SyEMkXJB8Yo/Vo+TKTo=
-github.com/google/pprof v0.0.0-20241210010833-40e02aabc2ad h1:a6HEuzUHeKH6hwfN/ZoQgRgVIWFJljSWa/zetS2WTvg=
+github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e h1:ijClszYn+mADRFY17kjQEVQ1XRhq2/JR1M3sGqeJoxs=
-github.com/google/pprof v0.0.0-20241210010833-40e02aabc2ad/go.mod h1:vavhavw2zAxS5dIdcRluK6cSGGPlZynqzFM8NdvU144=
+github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e/go.mod h1:boTsfXsheKC2y+lKOCMpSfarhxDeIzfZG1jqGcPl3cA=
 github.com/google/renameio v0.1.0/go.mod h1:KWCgfxg9yswjAJkECMjeO8J8rahYeXnNhOm40UhjYkI=
 github.com/google/s2a-go v0.1.7 h1:60BLSyTrOV4/haCDW4zb1guZItoSq8foHCXrAnjBo/o=
 github.com/google/s2a-go v0.1.7/go.mod h1:50CgR4k1jNlWBu4UfS4AcfhVe1r6pdZPygJ3R8F0Qdw=
@@ -597,8 +601,8 @@ github.com/jinzhu/gorm v0.0.0-20170222002820-5409931a1bb8/go.mod h1:Vla75njaFJ8c
 github.com/jinzhu/inflection v0.0.0-20170102125226-1c35d901db3d h1:jRQLvyVGL+iVtDElaEIDdKwpPqUIZJfzkNLV34htpEc=
 github.com/jinzhu/inflection v0.0.0-20170102125226-1c35d901db3d/go.mod h1:h+uFLlag+Qp1Va5pdKtLDYj+kHp5pxUVkryuEj+Srlc=
 github.com/jinzhu/now v1.1.1/go.mod h1:d3SSVoowX0Lcu0IBviAWJpolVfI5UJVZZ7cO71lE/z8=
-github.com/jmoiron/sqlx v1.3.3 h1:j82X0bf7oQ27XeqxicSZsTU5suPwKElg3oyxNn43iTk=
+github.com/jmoiron/sqlx v1.4.0 h1:1PLqN7S1UYp5t4SrVVnt4nUVNemrDAtxlulVe+Qgm3o=
-github.com/jmoiron/sqlx v1.3.3/go.mod h1:2BljVx/86SuTyjE+aPYlHCTNvZrnJXghYGpNiXLBMCQ=
+github.com/jmoiron/sqlx v1.4.0/go.mod h1:ZrZ7UsYB/weZdl2Bxg6jCRO9c3YHl8r3ahlKmRT4JLY=
 github.com/josharian/native v1.1.0 h1:uuaP0hAbW7Y4l0ZRQ6C9zfb7Mg1mbFKry/xzDAfmtLA=
 github.com/josharian/native v1.1.0/go.mod h1:7X/raswPFr05uY3HiLlYeyQntB6OO7E/d2Cu7qoaN2w=
 github.com/json-iterator/go v1.1.6/go.mod h1:+SdeFBvtyEkXs7REEP0seUULqWtbJapLOCVDaaPEHmU=
@@ -639,8 +643,9 @@ github.com/lib/pq v0.0.0-20150723085316-0dad96c0b94f/go.mod h1:5WUZQaWbwv1U+lTRe
 github.com/lib/pq v1.0.0/go.mod h1:5WUZQaWbwv1U+lTReE5YruASi9Al49XbQIvNi/34Woo=
 github.com/lib/pq v1.1.0/go.mod h1:5WUZQaWbwv1U+lTReE5YruASi9Al49XbQIvNi/34Woo=
 github.com/lib/pq v1.2.0/go.mod h1:5WUZQaWbwv1U+lTReE5YruASi9Al49XbQIvNi/34Woo=
-github.com/lib/pq v1.10.2 h1:AqzbZs4ZoCBp+GtejcpCpcxM3zlSMx29dXbUSeVtJb8=
 github.com/lib/pq v1.10.2/go.mod h1:AlVN5x4E4T544tWzH6hKfbfQvm3HdbOxrmggDNAPY9o=
+github.com/lib/pq v1.10.9 h1:YXG7RB+JIjhP29X+OtkiDnYaXQwpS4JEWq7dtCCRUEw=
+github.com/lib/pq v1.10.9/go.mod h1:AlVN5x4E4T544tWzH6hKfbfQvm3HdbOxrmggDNAPY9o=
 github.com/libdns/libdns v0.2.2 h1:O6ws7bAfRPaBsgAYt8MDe2HcNBGC29hkZ9MX2eUSX3s=
 github.com/libdns/libdns v0.2.2/go.mod h1:4Bj9+5CQiNMVGf87wjX4CY3HQJypUHRuLvlsfsZqLWQ=
 github.com/libp2p/go-buffer-pool v0.1.0 h1:oK4mSFcQz7cTQIfqbe4MIj9gLW+mnanjyFtc6cdF0Y8=
@@ -701,6 +706,8 @@ github.com/mattn/go-runewidth v0.0.16/go.mod h1:Jdepj2loyihRzMpdS35Xk/zdY8IAYHsh
 github.com/mattn/go-shellwords v1.0.12 h1:M2zGm7EW6UQJvDeQxo4T51eKPurbeFbe8WtebGE2xrk=
 github.com/mattn/go-shellwords v1.0.12/go.mod h1:EZzvwXDESEeg03EKmM+RmDnNOPKG4lLtQsUlTZDWQ8Y=
 github.com/mattn/go-sqlite3 v1.6.0/go.mod h1:FPy6KqzDD04eiIsT53CuJW3U88zkxoIYsOqkbpncsNc=
+github.com/mattn/go-sqlite3 v1.14.22 h1:2gZY6PC6kBnID23Tichd1K+Z0oS6nE/XwU+Vz/5o4kU=
+github.com/mattn/go-sqlite3 v1.14.22/go.mod h1:Uh1q+B4BYcTPb+yiD3kU8Ct7aC0hY9fxUwlHK0RXw+Y=
 github.com/matttproud/golang_protobuf_extensions v1.0.1/go.mod h1:D8He9yQNgCq6Z5Ld7szi9bcBfOoFv/3dc6xSMkL2PC0=
 github.com/mdlayher/genetlink v1.3.2 h1:KdrNKe+CTu+IbZnm/GVUMXSqBBLqcGpRDa0xkQy56gw=
 github.com/mdlayher/genetlink v1.3.2/go.mod h1:tcC3pkCrPUGIKKsCsp0B3AdaaKuHtaxoJRz3cc+528o=
@@ -714,8 +721,8 @@ github.com/mholt/acmez/v2 v2.0.3 h1:CgDBlEwg3QBp6s45tPQmFIBrkRIkBT4rW4orMM6p4sw=
 github.com/mholt/acmez/v2 v2.0.3/go.mod h1:pQ1ysaDeGrIMvJ9dfJMk5kJNkn7L2sb3UhyrX6Q91cw=
 github.com/miekg/dns v1.1.26/go.mod h1:bPDLeHnStXmXAq1m/Ch/hvfNHr14JKNPMBo3VZKjuso=
 github.com/miekg/dns v1.1.41/go.mod h1:p6aan82bvRIyn+zDIv9xYNUpwa73JcSh9BKwknJysuI=
-github.com/miekg/dns v1.1.62 h1:cN8OuEF1/x5Rq6Np+h1epln8OiyPWV+lROx9LxcGgIQ=
+github.com/miekg/dns v1.1.65 h1:0+tIPHzUW0GCge7IiK3guGP57VAw7hoPDfApjkMD1Fc=
-github.com/miekg/dns v1.1.62/go.mod h1:mvDlcItzm+br7MToIKqkglaGhlFMHJ9DTNNWONWXbNQ=
+github.com/miekg/dns v1.1.65/go.mod h1:Dzw9769uoKVaLuODMDZz9M6ynFU6Em65csPuoi8G0ck=
 github.com/miekg/pkcs11 v1.0.2/go.mod h1:XsNlhZGX73bx86s2hdc/FuaLm2CPZJemRLMA+WTFxgs=
 github.com/miekg/pkcs11 v1.1.1 h1:Ugu9pdy6vAYku5DEpVWVFPYnzV+bxB+iRdbuFSu7TvU=
 github.com/miekg/pkcs11 v1.1.1/go.mod h1:XsNlhZGX73bx86s2hdc/FuaLm2CPZJemRLMA+WTFxgs=
@@ -809,6 +816,8 @@ github.com/multiformats/go-varint v0.0.7/go.mod h1:r8PUYw/fD/SjBCiKOoDlGF6QawOEL
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 h1:C3w9PqII01/Oq1c1nUAm88MOHcQC9l5mIlSMApZMrHA=
 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822/go.mod h1:+n7T8mK8HuQTcFwEeznm/DIxMOiR9yIdICNftLE1DvQ=
 github.com/mwitkow/go-conntrack v0.0.0-20161129095857-cc309e4a2223/go.mod h1:qRWi+5nqEBWmkhHvq77mSJWrCKwh8bxhgT7d/eI7P4U=
+github.com/ncruces/go-strftime v0.1.9 h1:bY0MQC28UADQmHmaF5dgpLmImcShSi2kHU9XLdhx/f4=
+github.com/ncruces/go-strftime v0.1.9/go.mod h1:Fwc5htZGVVkseilnfgOVb9mKy6w1naJmn9CehxcKcls=
 github.com/niemeyer/pretty v0.0.0-20200227124842-a10e7caefd8e/go.mod h1:zD1mROLANZcx1PVRCS0qkT7pwLkGfwJo4zjcN/Tysno=
 github.com/onsi/ginkgo v1.6.0/go.mod h1:lLunBs/Ym6LB5Z9jYTR76FiuTmxDTDusOGeTQH+WWjE=
 github.com/onsi/ginkgo v1.12.0/go.mod h1:oUhWkIvk5aDxtKvDDuw8gItl8pKl42LzjC9KZE0HfGg=
@@ -925,6 +934,8 @@ github.com/quic-go/webtransport-go v0.8.0 h1:HxSrwun11U+LlmwpgM1kEqIqH90IT4N8auv
 github.com/quic-go/webtransport-go v0.8.0/go.mod h1:N99tjprW432Ut5ONql/aUhSLT0YVSlwHohQsuac9WaM=
 github.com/raulk/go-watchdog v1.3.0 h1:oUmdlHxdkXRJlwfG0O9omj8ukerm8MEQavSiDTEtBsk=
 github.com/raulk/go-watchdog v1.3.0/go.mod h1:fIvOnLbF0b0ZwkB9YU4mOW9Did//4vPZtDqv66NfsMU=
+github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
+github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
 github.com/rivo/uniseg v0.2.0/go.mod h1:J6wj4VEh+S6ZtnVlnTBMWIodfgj8LQOQFoIToxlJtxc=
 github.com/rivo/uniseg v0.4.7 h1:WUdvkW8uEhrYfLC4ZzdpI2ztxP1I582+49Oc5Mq64VQ=
 github.com/rivo/uniseg v0.4.7/go.mod h1:FN3SvrM+Zdj16jyLfmOkMNblXMcoc8DfTHruCPUcx88=
@@ -1069,6 +1080,13 @@ github.com/warpfork/go-wish v0.0.0-20220906213052-39a1cc7a02d0 h1:GDDkbFiaK8jsSD
 github.com/warpfork/go-wish v0.0.0-20220906213052-39a1cc7a02d0/go.mod h1:x6AKhvSSexNrVSrViXSHUEbICjmGXhtgABaHIySUSGw=
 github.com/weppos/publicsuffix-go v0.15.1-0.20210511084619-b1f36a2d6c0b h1:FsyNrX12e5BkplJq7wKOLk0+C6LZ+KGXvuEcKUYm5ss=
 github.com/weppos/publicsuffix-go v0.15.1-0.20210511084619-b1f36a2d6c0b/go.mod h1:HYux0V0Zi04bHNwOHy4cXJVz/TQjYonnF6aoYhj+3QE=
+github.com/xeipuuv/gojsonpointer v0.0.0-20180127040702-4e3ac2762d5f/go.mod h1:N2zxlSyiKSe5eX1tZViRH5QA0qijqEDrYZiPEAiq3wU=
+github.com/xeipuuv/gojsonpointer v0.0.0-20190905194746-02993c407bfb h1:zGWFAtiMcyryUHoUjUJX0/lt1H2+i2Ka2n+D3DImSNo=
+github.com/xeipuuv/gojsonpointer v0.0.0-20190905194746-02993c407bfb/go.mod h1:N2zxlSyiKSe5eX1tZViRH5QA0qijqEDrYZiPEAiq3wU=
+github.com/xeipuuv/gojsonreference v0.0.0-20180127040603-bd5ef7bd5415 h1:EzJWgHovont7NscjpAxXsDA8S8BMYve8Y5+7cuRE7R0=
+github.com/xeipuuv/gojsonreference v0.0.0-20180127040603-bd5ef7bd5415/go.mod h1:GwrjFmJcFw6At/Gs6z4yjiIwzuJ1/+UwLxMQDVQXShQ=
+github.com/xeipuuv/gojsonschema v1.2.0 h1:LhYJRs+L4fBtjZUfuSZIKGeVu0QRy8e5Xi7D17UxZ74=
+github.com/xeipuuv/gojsonschema v1.2.0/go.mod h1:anYRn/JVcOK2ZgGU+IjEV4nwlhoK5sQluxsYJ78Id3Y=
 github.com/xordataexchange/crypt v0.0.3-0.20170626215501-b2862e3d0a77/go.mod h1:aYKd//L2LvnjZzWKhF00oedf4jCCReLcmhLdhm1A27Q=
 github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.2.1/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
@@ -1178,8 +1196,8 @@ golang.org/x/crypto v0.0.0-20210711020723-a769d52b0f97/go.mod h1:GvvjBRRGRdwPK5y
 golang.org/x/crypto v0.0.0-20210921155107-089bfa567519/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc=
 golang.org/x/crypto v0.3.0/go.mod h1:hebNnKkNXi2UzZN1eVRvBB7co0a+JxK6XbPiWVs/3J4=
 golang.org/x/crypto v0.19.0/go.mod h1:Iy9bg/ha4yyC70EfRS8jz+B6ybOBKMaSxLj6P6oBDfU=
-golang.org/x/crypto v0.32.0 h1:euUpcYgM8WcP71gNpTqQCn6rC2t6ULUPiOzfWaXVVfc=
+golang.org/x/crypto v0.33.0 h1:IOBPskki6Lysi0lo9qQvbxiQ+FvsCC/YWOecCHAixus=
-golang.org/x/crypto v0.32.0/go.mod h1:ZnnJkOaASj8g0AjIduWNlq2NRxL0PlBrbKVyZ6V/Ugc=
+golang.org/x/crypto v0.33.0/go.mod h1:bVdXmD7IV/4GdElGPozy6U7lWdRXA4qyRVGJV57uQ5M=
 golang.org/x/crypto/x509roots/fallback v0.0.0-20240507223354-67b13616a595 h1:TgSqweA595vD0Zt86JzLv3Pb/syKg8gd5KMGGbJPYFw=
 golang.org/x/crypto/x509roots/fallback v0.0.0-20240507223354-67b13616a595/go.mod h1:kNa9WdvYnzFwC79zRpLRMJbdEFlhyM5RPFBBZp/wWH8=
 golang.org/x/exp v0.0.0-20190121172915-509febef88a4/go.mod h1:CJ0aWSM057203Lf6IL+f9T1iT9GByDxfZKAQTCR3kQA=
@@ -1196,8 +1214,8 @@ golang.org/x/mod v0.3.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
 golang.org/x/mod v0.4.2/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
 golang.org/x/mod v0.6.0-dev.0.20220419223038-86c51ed26bb4/go.mod h1:jJ57K6gSWd91VN4djpZkiMVwK6gcyfeH4XE8wZrZaV4=
 golang.org/x/mod v0.8.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
-golang.org/x/mod v0.22.0 h1:D4nJWe9zXqHOmWqj4VMOJhvzj7bEZg4wEYa759z1pH4=
+golang.org/x/mod v0.23.0 h1:Zb7khfcRGKk+kqfxFaP5tZqCnDZMjC5VtUBs87Hr6QM=
-golang.org/x/mod v0.22.0/go.mod h1:6SkKJ3Xj0I0BrPOZoBy3bdMptDDU9oJrpohJ3eWZ1fY=
+golang.org/x/mod v0.23.0/go.mod h1:6SkKJ3Xj0I0BrPOZoBy3bdMptDDU9oJrpohJ3eWZ1fY=
 golang.org/x/net v0.0.0-20180724234803-3673e40ba225/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20180826012351-8a410e7b638d/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20180906233101-161cd47e91fd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
@@ -1219,8 +1237,8 @@ golang.org/x/net v0.0.0-20220722155237-a158d28d115b/go.mod h1:XRhObCWvk6IyKnWLug
 golang.org/x/net v0.2.0/go.mod h1:KqCZLdyyvdV855qA2rE3GC2aiw5xGR5TEjj8smXukLY=
 golang.org/x/net v0.6.0/go.mod h1:2Tu9+aMcznHK/AK1HMvgo6xiTLG5rD5rZLDS+rp2Bjs=
 golang.org/x/net v0.10.0/go.mod h1:0qNGK6F8kojg2nk9dLZ2mShWaEBan6FAoqfSigmmuDg=
-golang.org/x/net v0.34.0 h1:Mb7Mrk043xzHgnRM88suvJFwzVrRfHEHJEl5/71CKw0=
+golang.org/x/net v0.35.0 h1:T5GQRQb2y08kTAByq9L4/bz8cipCdA8FbRTXewonqY8=
-golang.org/x/net v0.34.0/go.mod h1:di0qlW3YNM5oh6GqDGQr92MyTozJPmybPK4Ev/Gm31k=
+golang.org/x/net v0.35.0/go.mod h1:EglIi67kWsHKlRzzVMUD93VMSWGFOMSZgxFjparz1Qk=
 golang.org/x/oauth2 v0.0.0-20180821212333-d2e6202438be/go.mod h1:N/0e6XlmueqKjAGxoOufVs8QHGRruUQn6yWY3a++T0U=
 golang.org/x/oauth2 v0.24.0 h1:KTBBxWqUa0ykRPLtV69rRto9TLXcqYkeswu48x/gvNE=
 golang.org/x/oauth2 v0.24.0/go.mod h1:XYTD2NtWslqkgxebSiOHnXEap4TF09sJSc7H1sXbhtI=
@@ -1234,8 +1252,8 @@ golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJ
 golang.org/x/sync v0.0.0-20210220032951-036812b2e83c/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20220722155255-886fb9371eb4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.1.0/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.10.0 h1:3NQrjDixjgGwUOCaF8w2+VYHv0Ve/vGYSbdkTa98gmQ=
+golang.org/x/sync v0.11.0 h1:GGz8+XQP4FvTTrjZPzNKTMFtSXH80RAzG+5ghFPgK9w=
-golang.org/x/sync v0.10.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
+golang.org/x/sync v0.11.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
 golang.org/x/sys v0.0.0-20180823144017-11551d06cbcc/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20180830151530-49385e6e1522/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20180905080454-ebe1bf3edb33/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
@@ -1283,8 +1301,8 @@ golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.8.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.17.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/sys v0.29.0 h1:TPYlXGxvx1MGTn2GiZDhnjPA9wZzZeGKHHmKhHYvgaU=
+golang.org/x/sys v0.31.0 h1:ioabZlmFYtWhL+TRYpcnNlLwhyxaM9kWTDEmfnprqik=
-golang.org/x/sys v0.29.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
+golang.org/x/sys v0.31.0/go.mod h1:BJP2sWEmIv4KK5OTEluFJCKSidICx8ciO85XgH3Ak8k=
 golang.org/x/term v0.0.0-20201117132131-f5c789dd3221/go.mod h1:Nr5EML6q2oocZ2LXRh80K7BxOlk5/8JxuGnuhpl+muw=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
 golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8=
@@ -1292,8 +1310,8 @@ golang.org/x/term v0.2.0/go.mod h1:TVmDHMZPmdnySmBfhjOoOdhjzdE1h4u1VwSiw2l1Nuc=
 golang.org/x/term v0.5.0/go.mod h1:jMB1sMXY+tzblOD4FWmEbocvup2/aLOaQEp7JmGp78k=
 golang.org/x/term v0.8.0/go.mod h1:xPskH00ivmX89bAKVGSKKtLOWNx2+17Eiy94tnKShWo=
 golang.org/x/term v0.17.0/go.mod h1:lLRBjIVuehSbZlaOtGMbcMncT+aqLLLmKrsjNrUguwk=
-golang.org/x/term v0.28.0 h1:/Ts8HFuMR2E6IP/jlo7QVLZHggjKQbhu/7H0LJFr3Gg=
+golang.org/x/term v0.29.0 h1:L6pJp37ocefwRRtYPKSWOWzOtWSxVajvz2ldH/xi3iU=
-golang.org/x/term v0.28.0/go.mod h1:Sw/lC2IAUZ92udQNf3WodGtn4k/XoLyZoh8v/8uiwek=
+golang.org/x/term v0.29.0/go.mod h1:6bl4lRlvVuDgSf3179VpIxBF0o10JUpXWOnI7nErv7s=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
@@ -1304,8 +1322,8 @@ golang.org/x/text v0.4.0/go.mod h1:mrYo+phRRbMaCq/xk9113O4dZlRixOauAjOtrjsXDZ8=
 golang.org/x/text v0.7.0/go.mod h1:mrYo+phRRbMaCq/xk9113O4dZlRixOauAjOtrjsXDZ8=
 golang.org/x/text v0.9.0/go.mod h1:e1OnstbJyHTd6l/uOt8jFFHp6TRDWZR/bV3emEE/zU8=
 golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
-golang.org/x/text v0.21.0 h1:zyQAAkrwaneQ066sspRyJaG9VNi/YJ1NfzcGB3hZ/qo=
+golang.org/x/text v0.22.0 h1:bofq7m3/HAFvbF51jz3Q9wLg3jkvSPuiZu/pD1XwgtM=
-golang.org/x/text v0.21.0/go.mod h1:4IBbMaMmOPCJ8SecivzSH54+73PCFmPWxNTLm+vZkEQ=
+golang.org/x/text v0.22.0/go.mod h1:YRoo4H8PVmsu+E3Ou7cqLVH8oXWIHVoX0jqUWALQhfY=
 golang.org/x/time v0.8.0 h1:9i3RxcPv3PZnitoVGMPDKZSq1xW1gK1Xy3ArNOGZfEg=
 golang.org/x/time v0.8.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
@@ -1327,8 +1345,8 @@ golang.org/x/tools v0.0.0-20210106214847-113979e3529a/go.mod h1:emZCQorbCU4vsT4f
 golang.org/x/tools v0.1.5/go.mod h1:o0xws9oXOQQZyjljx8fwUC0k7L1pTE6eaCbjGeHmOkk=
 golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc=
 golang.org/x/tools v0.6.0/go.mod h1:Xwgl3UAJ/d3gWutnCtw505GrjyAbvKui8lOU390QaIU=
-golang.org/x/tools v0.29.0 h1:Xx0h3TtM9rzQpQuR4dKLrdglAmCEN5Oi+P74JdhdzXE=
+golang.org/x/tools v0.30.0 h1:BgcpHewrV5AUp2G9MebG4XPFI1E2W41zU1SaqVA9vJY=
-golang.org/x/tools v0.29.0/go.mod h1:KMQVMRsVxU6nHCFXrBPhDB8XncLNLM0lIy/F14RP588=
+golang.org/x/tools v0.30.0/go.mod h1:c347cR/OJfw5TI+GfX7RUPNMdDRRbjvYTS0jPyvsVtY=
 golang.org/x/xerrors v0.0.0-20190410155217-1f06c39b4373/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20190513163551-3ee3066db522/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
@@ -1420,3 +1438,27 @@ k8s.io/klog/v2 v2.110.1 h1:U/Af64HJf7FcwMcXyKm2RPM22WZzyR7OSpYj5tg3cL0=
 k8s.io/klog/v2 v2.110.1/go.mod h1:YGtd1984u+GgbuZ7e08/yBuAfKLSO0+uR1Fhi6ExXjo=
 lukechampine.com/blake3 v1.3.0 h1:sJ3XhFINmHSrYCgl958hscfIa3bw8x4DqMP3u1YvoYE=
 lukechampine.com/blake3 v1.3.0/go.mod h1:0OFRp7fBtAylGVCO40o87sbupkyIGgbpv1+M1k1LM6k=
+modernc.org/cc/v4 v4.24.4 h1:TFkx1s6dCkQpd6dKurBNmpo+G8Zl4Sq/ztJ+2+DEsh0=
+modernc.org/cc/v4 v4.24.4/go.mod h1:uVtb5OGqUKpoLWhqwNQo/8LwvoiEBLvZXIQ/SmO6mL0=
+modernc.org/ccgo/v4 v4.23.16 h1:Z2N+kk38b7SfySC1ZkpGLN2vthNJP1+ZzGZIlH7uBxo=
+modernc.org/ccgo/v4 v4.23.16/go.mod h1:nNma8goMTY7aQZQNTyN9AIoJfxav4nvTnvKThAeMDdo=
+modernc.org/fileutil v1.3.0 h1:gQ5SIzK3H9kdfai/5x41oQiKValumqNTDXMvKo62HvE=
+modernc.org/fileutil v1.3.0/go.mod h1:XatxS8fZi3pS8/hKG2GH/ArUogfxjpEKs3Ku3aK4JyQ=
+modernc.org/gc/v2 v2.6.3 h1:aJVhcqAte49LF+mGveZ5KPlsp4tdGdAOT4sipJXADjw=
+modernc.org/gc/v2 v2.6.3/go.mod h1:YgIahr1ypgfe7chRuJi2gD7DBQiKSLMPgBQe9oIiito=
+modernc.org/libc v1.61.13 h1:3LRd6ZO1ezsFiX1y+bHd1ipyEHIJKvuprv0sLTBwLW8=
+modernc.org/libc v1.61.13/go.mod h1:8F/uJWL/3nNil0Lgt1Dpz+GgkApWh04N3el3hxJcA6E=
+modernc.org/mathutil v1.7.1 h1:GCZVGXdaN8gTqB1Mf/usp1Y/hSqgI2vAGGP4jZMCxOU=
+modernc.org/mathutil v1.7.1/go.mod h1:4p5IwJITfppl0G4sUEDtCr4DthTaT47/N3aT6MhfgJg=
+modernc.org/memory v1.8.2 h1:cL9L4bcoAObu4NkxOlKWBWtNHIsnnACGF/TbqQ6sbcI=
+modernc.org/memory v1.8.2/go.mod h1:ZbjSvMO5NQ1A2i3bWeDiVMxIorXwdClKE/0SZ+BMotU=
+modernc.org/opt v0.1.4 h1:2kNGMRiUjrp4LcaPuLY2PzUfqM/w9N23quVwhKt5Qm8=
+modernc.org/opt v0.1.4/go.mod h1:03fq9lsNfvkYSfxrfUhZCWPk1lm4cq4N+Bh//bEtgns=
+modernc.org/sortutil v1.2.1 h1:+xyoGf15mM3NMlPDnFqrteY07klSFxLElE2PVuWIJ7w=
+modernc.org/sortutil v1.2.1/go.mod h1:7ZI3a3REbai7gzCLcotuw9AC4VZVpYMjDzETGsSMqJE=
+modernc.org/sqlite v1.36.3 h1:qYMYlFR+rtLDUzuXoST1SDIdEPbX8xzuhdF90WsX1ss=
+modernc.org/sqlite v1.36.3/go.mod h1:ADySlx7K4FdY5MaJcEv86hTJ0PjedAloTUuif0YS3ws=
+modernc.org/strutil v1.2.1 h1:UneZBkQA+DX2Rp35KcM69cSsNES9ly8mQWD71HKlOA0=
+modernc.org/strutil v1.2.1/go.mod h1:EHkiggD70koQxjVdSBM3JKM7k6L0FbGE5eymy9i3B9A=
+modernc.org/token v1.1.0 h1:Xl7Ap9dKaEs5kLoOQeQmPWevfnk/DM5qcLcYlA8ys6Y=
+modernc.org/token v1.1.0/go.mod h1:UGzOrNV1mAFSEB63lOFHIpNRUVMvYTc6yu1SMY/XTDM=

internal/api/service.go

@@ -1,165 +0,0 @@
-package api
-
-import (
-	"encoding/json"
-	"fmt"
-	"github.com/distribution/reference"
-	"maps"
-	"reflect"
-	"regexp"
-	"slices"
-	"uncloud/internal/machine/api/pb"
-)
-
-const (
-	ServiceModeReplicated = "replicated"
-	ServiceModeGlobal     = "global"
-)
-
-type ServiceSpec struct {
-	Container ContainerSpec
-	// Mode is the replication mode of the service. Default is ServiceModeReplicated if empty.
-	Mode string
-	Name string
-	// Ports defines what service ports to publish to make the service accessible outside the cluster.
-	Ports []PortSpec
-	// Replicas is the number of containers to run for the service. Only valid for a replicated service.
-	Replicas uint
-}
-
-func (s *ServiceSpec) Validate() error {
-	if err := s.Container.Validate(); err != nil {
-		return err
-	}
-
-	switch s.Mode {
-	case "", ServiceModeGlobal, ServiceModeReplicated:
-	default:
-		return fmt.Errorf("invalid mode: %q", s.Mode)
-	}
-
-	// TODO: validate there is no conflict between ports.
-	// TODO: return error if there are non-HTTP/HTTPS ingress ports that we don't support yet.
-
-	return nil
-}
-
-// Equals returns true if the service spec is equal to the given spec ignoring the number of replicas.
-func (s *ServiceSpec) Equals(spec ServiceSpec) bool {
-	// TODO: ignore order of ports.
-	sCopy := *s
-	// Ignore the number of replicas when comparing.
-	sCopy.Replicas = 0
-	spec.Replicas = 0
-	return reflect.DeepEqual(*s, spec)
-}
-
-var serviceIDRegexp = regexp.MustCompile("^[0-9a-f]{32}$")
-
-func ValidateServiceID(id string) bool {
-	return serviceIDRegexp.MatchString(id)
-}
-
-type ContainerSpec struct {
-	Command []string
-	Image   string
-	// Run a custom init inside the container. If nil, use the daemon's configured settings.
-	Init *bool
-	// List of volumes to bind mount into the container.
-	Volumes []string
-}
-
-func (s *ContainerSpec) Validate() error {
-	if _, err := reference.ParseDockerRef(s.Image); err != nil {
-		return fmt.Errorf("invalid image: %w", err)
-	}
-
-	return nil
-}
-
-type Service struct {
-	ID         string
-	Name       string
-	Mode       string
-	Containers []MachineContainer
-}
-
-type MachineContainer struct {
-	MachineID string
-	Container Container
-}
-
-// Endpoints returns the exposed HTTP and HTTPS endpoints of the service.
-func (s *Service) Endpoints() []string {
-	endpoints := make(map[string]struct{})
-
-	// Container specs may differ between containers in the same service, e.g. during a rolling update,
-	// so we need to collect all unique endpoints.
-	for _, ctr := range s.Containers {
-		ports, err := ctr.Container.ServicePorts()
-		if err != nil {
-			continue
-		}
-
-		for _, port := range ports {
-			protocol := ""
-			switch port.Protocol {
-			case ProtocolHTTP:
-				protocol = "http"
-			case ProtocolHTTPS:
-				protocol = "https"
-			default:
-				continue
-			}
-
-			if port.Hostname == "" {
-				// There shouldn't be http(s) ports without a hostname but just in case ignore them.
-				continue
-			}
-
-			endpoint := fmt.Sprintf("%s://%s", protocol, port.Hostname)
-			if port.PublishedPort != 0 {
-				// For non-standard ports (80/443), include the port in the URL.
-				if !(port.Protocol == ProtocolHTTP && port.PublishedPort == 80) &&
-					!(port.Protocol == ProtocolHTTPS && port.PublishedPort == 443) {
-					endpoint += fmt.Sprintf(":%d", port.PublishedPort)
-				}
-			}
-
-			endpoint += fmt.Sprintf(" → :%d", port.ContainerPort)
-			endpoints[endpoint] = struct{}{}
-		}
-	}
-
-	return slices.Sorted(maps.Keys(endpoints))
-}
-
-func ServiceFromProto(s *pb.Service) (Service, error) {
-	var err error
-	containers := make([]MachineContainer, len(s.Containers))
-	for i, sc := range s.Containers {
-		containers[i], err = machineContainerFromProto(sc)
-		if err != nil {
-			return Service{}, err
-		}
-	}
-
-	return Service{
-		ID:         s.Id,
-		Name:       s.Name,
-		Mode:       s.Mode,
-		Containers: containers,
-	}, nil
-}
-
-func machineContainerFromProto(sc *pb.Service_Container) (MachineContainer, error) {
-	var c Container
-	if err := json.Unmarshal(sc.Container, &c); err != nil {
-		return MachineContainer{}, fmt.Errorf("unmarshal container: %w", err)
-	}
-
-	return MachineContainer{
-		MachineID: sc.MachineId,
-		Container: c,
-	}, nil
-}

internal/cli/build.go

@@ -0,0 +1,176 @@
+package cli
+
+import (
+	"context"
+	"encoding/base64"
+	"encoding/json"
+	"fmt"
+	"os"
+
+	composetypes "github.com/compose-spec/compose-go/v2/types"
+	"github.com/distribution/reference"
+	"github.com/docker/cli/cli/config"
+	"github.com/docker/docker/api/types"
+	"github.com/docker/docker/api/types/image"
+	dockerclient "github.com/docker/docker/client"
+	"github.com/docker/docker/pkg/archive"
+	"github.com/docker/docker/pkg/jsonmessage"
+	"github.com/docker/docker/registry"
+	"github.com/moby/term"
+)
+
+type BuildOptions struct {
+	Files    []string
+	Profiles []string
+	Services []string
+	Push     bool
+	NoCache  bool
+}
+
+// GetServicesThatNeedBuild returns a map of services that require building
+func GetServicesThatNeedBuild(project *composetypes.Project) map[string]composetypes.ServiceConfig {
+	servicesToBuild := make(map[string]composetypes.ServiceConfig, len(project.Services))
+	for serviceName, service := range project.Services {
+		if service.Build == nil {
+			continue
+		}
+		servicesToBuild[serviceName] = service
+	}
+	return servicesToBuild
+}
+
+// BuildServices builds the services defined in the provided map.
+func BuildServices(ctx context.Context, servicesToBuild map[string]composetypes.ServiceConfig, opts BuildOptions) error {
+	fmt.Println("Building services...")
+
+	// Init docker client (can be local or remote, depending on DOCKER_HOST environment variable)
+	dockerCli, err := dockerclient.NewClientWithOpts(dockerclient.FromEnv, dockerclient.WithAPIVersionNegotiation())
+	if err != nil {
+		return err
+	}
+	defer dockerCli.Close()
+
+	serviceImages := make(map[string]string, len(servicesToBuild))
+
+	// Build the services using the local docker client and compose libraries
+	for _, service := range servicesToBuild {
+		fmt.Printf("Building service: %s\n", service.Name)
+		imageName, err := buildSingleService(ctx, dockerCli, service, opts)
+		if err != nil {
+			return fmt.Errorf("build service %s: %w", service.Name, err)
+		}
+		serviceImages[service.Name] = imageName
+	}
+	fmt.Printf("Service images are built.\n")
+
+	if opts.Push {
+		err = pushServiceImages(ctx, dockerCli, serviceImages)
+	}
+
+	return err
+}
+
+// buildSingleService builds a single service using the Docker client and Compose libraries.
+func buildSingleService(ctx context.Context, dockerCli *dockerclient.Client, service composetypes.ServiceConfig, opts BuildOptions) (string, error) {
+	if service.Build == nil {
+		return "", fmt.Errorf("service %s has no build configuration", service.Name)
+	}
+	if service.Image == "" {
+		return "", fmt.Errorf("service %s has no image specified; building services without image is not supported yet", service.Name)
+	}
+
+	buildContextPath := service.Build.Context
+	imageName := service.Image
+
+	// Create a tar archive of the build context
+	buildContext, err := archive.TarWithOptions(buildContextPath, &archive.TarOptions{})
+	if err != nil {
+		return "", fmt.Errorf("failed to create build context for service %s: %w", service.Name, err)
+	}
+
+	buildOptions := types.ImageBuildOptions{
+		// TODO: Support Dockerfiles outside the build context
+		// See https://github.com/docker/compose/blob/cf89fd1aa1328d5af77658ccc5a1e1b29981ae80/pkg/compose/build_classic.go#L92
+		Dockerfile: service.Build.Dockerfile,
+		Tags:       []string{imageName},
+		Remove:     true, // Remove intermediate containers
+		NoCache:    opts.NoCache,
+	}
+
+	buildResponse, err := dockerCli.ImageBuild(ctx, buildContext, buildOptions)
+	if err != nil {
+		return "", fmt.Errorf("failed to build image for service %s: %w", service.Name, err)
+	}
+	defer buildResponse.Body.Close()
+
+	// Display the build response
+	fd, isTerminal := term.GetFdInfo(os.Stdout)
+	if err := jsonmessage.DisplayJSONMessagesStream(buildResponse.Body, os.Stdout, fd, isTerminal, nil); err != nil {
+		return "", fmt.Errorf("failed to display build response for service %s: %w", service.Name, err)
+	}
+
+	return imageName, nil
+}
+
+// pushSingleServiceImage pushes a single service image.
+func pushSingleServiceImage(ctx context.Context, dockerCli *dockerclient.Client, serviceName string, imageName string) error {
+	ref, err := reference.ParseNormalizedNamed(imageName)
+	if err != nil {
+		return err
+	}
+
+	repoInfo, err := registry.ParseRepositoryInfo(ref)
+	if err != nil {
+		return err
+	}
+
+	registryKey := repoInfo.Index.Name
+	if repoInfo.Index.Official {
+		registryKey = registry.IndexServer
+	}
+
+	// Load the Docker config file with auth details, if available
+	configFile := config.LoadDefaultConfigFile(os.Stderr)
+
+	authConfig, err := configFile.GetAuthConfig(registryKey)
+	if err != nil {
+		return err
+	}
+
+	authJSON, err := json.Marshal(authConfig)
+	if err != nil {
+		return fmt.Errorf("failed to marshal auth config for registry %s: %w", registryKey, err)
+	}
+
+	authStr := base64.URLEncoding.EncodeToString(authJSON)
+
+	pushOptions := image.PushOptions{
+		RegistryAuth: authStr,
+	}
+	pushResponse, err := dockerCli.ImagePush(ctx, imageName, pushOptions)
+	if err != nil {
+		return fmt.Errorf("failed to push image %s: %w", imageName, err)
+	}
+	defer pushResponse.Close()
+
+	fmt.Printf("Pushing image %s for service %s...\n", imageName, serviceName)
+
+	fd, isTerminal := term.GetFdInfo(os.Stdout)
+	if err := jsonmessage.DisplayJSONMessagesStream(pushResponse, os.Stdout, fd, isTerminal, nil); err != nil {
+		return fmt.Errorf("failed to display push response for image %s: %w", imageName, err)
+	}
+
+	fmt.Printf("Image %s pushed successfully.\n", imageName)
+	return nil
+}
+
+// pushServiceImages pushes all built service images to the registry.
+func pushServiceImages(ctx context.Context, dockerCli *dockerclient.Client, serviceImages map[string]string) error {
+	fmt.Printf("Pushing images...\n")
+	for serviceName, imageName := range serviceImages {
+		if err := pushSingleServiceImage(ctx, dockerCli, serviceName, imageName); err != nil {
+			return fmt.Errorf("push image for service %s: %w", serviceName, err)
+		}
+	}
+	return nil
+}

internal/cli/cli.go

@@ -4,90 +4,125 @@ import (
 	"context"
 	"errors"
 	"fmt"
-	"github.com/docker/cli/cli/streams"
 	"net/netip"
 	"os"
-	"uncloud/internal/cli/client"
-	"uncloud/internal/cli/client/connector"
-	"uncloud/internal/cli/config"
-	"uncloud/internal/fs"
-	"uncloud/internal/machine"
-	"uncloud/internal/machine/api/pb"
-	"uncloud/internal/sshexec"
 
 	"github.com/charmbracelet/huh"
+	"github.com/docker/cli/cli/streams"
+	"github.com/psviderski/uncloud/internal/cli/config"
+	"github.com/psviderski/uncloud/internal/fs"
+	"github.com/psviderski/uncloud/internal/machine"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/internal/sshexec"
+	"github.com/psviderski/uncloud/pkg/api"
+	"github.com/psviderski/uncloud/pkg/client"
+	"github.com/psviderski/uncloud/pkg/client/connector"
+	"google.golang.org/grpc/codes"
+	"google.golang.org/grpc/status"
 	"google.golang.org/protobuf/types/known/emptypb"
 )
 
-const defaultClusterName = "default"
+const defaultContextName = "default"
 
 type CLI struct {
-	config *config.Config
+	Config *config.Config
+	conn   *config.MachineConnection
 }
 
-func New(configPath string) (*CLI, error) {
+// New creates a new CLI instance with the given config path or remote machine connection.
+// If the connection is provided, the config is ignored for all operations which is useful for interacting with
+// a cluster without creating a config.
+func New(configPath string, conn *config.MachineConnection) (*CLI, error) {
+	if conn != nil {
+		return &CLI{conn: conn}, nil
+	}
+
 	cfg, err := config.NewFromFile(configPath)
 	if err != nil {
 		return nil, fmt.Errorf("read Uncloud config: %w", err)
 	}
+
 	return &CLI{
-		config: cfg,
+		Config: cfg,
 	}, nil
 }
 
-func (cli *CLI) CreateCluster(name string) error {
+func (cli *CLI) CreateContext(name string) error {
-	if _, ok := cli.config.Clusters[name]; ok {
+	if _, ok := cli.Config.Contexts[name]; ok {
-		return fmt.Errorf("cluster %q already exists", name)
+		return fmt.Errorf("context '%s' already exists", name)
 	}
-	cli.config.Clusters[name] = &config.Cluster{
+	cli.Config.Contexts[name] = &config.Context{
 		Name: name,
 	}
-	return cli.config.Save()
+	return cli.Config.Save()
 }
 
-func (cli *CLI) SetCurrentCluster(name string) error {
+func (cli *CLI) SetCurrentContext(name string) error {
-	if _, ok := cli.config.Clusters[name]; !ok {
+	if _, ok := cli.Config.Contexts[name]; !ok {
-		return client.ErrNotFound
+		return api.ErrNotFound
 	}
-	cli.config.CurrentCluster = name
+	cli.Config.CurrentContext = name
-	return cli.config.Save()
+	return cli.Config.Save()
 }
 
-func (cli *CLI) ConnectCluster(ctx context.Context, clusterName string) (*client.Client, error) {
+// ConnectCluster connects to a cluster using the given context name or the current context if not specified.
-	if len(cli.config.Clusters) == 0 {
+// If the CLI was initialised with a machine connection, the config is ignored and the connection is used instead.
-		return nil, errors.New(
+func (cli *CLI) ConnectCluster(ctx context.Context, contextName string) (*client.Client, error) {
-			"no clusters found in the Uncloud config. " +
+	if cli.conn != nil {
-				"Please initialise a cluster with `uncloud machine init` first",
+		return connectCluster(ctx, *cli.conn)
+	}
+
+	if len(cli.Config.Contexts) == 0 {
+		return nil, fmt.Errorf(
+			"no cluster contexts found in the Uncloud config (%s). "+
+				"Please initialise a cluster with 'uncloud machine init' first",
+			cli.Config.Path(),
 		)
 	}
-	if clusterName == "" {
+	if contextName == "" {
 		// If the cluster is not specified, use the current cluster if set.
-		if cli.config.CurrentCluster == "" {
+		if cli.Config.CurrentContext == "" {
-			return nil, errors.New(
-				"the current cluster is not set in the Uncloud config. " +
-					"Please specify a cluster with the --cluster flag or set current_cluster in the config",
-			)
-		}
-		if _, ok := cli.config.Clusters[cli.config.CurrentCluster]; !ok {
 			return nil, fmt.Errorf(
-				"current cluster %q not found in the config. "+
+				"the current cluster context is not set in the Uncloud config (%s). "+
-					"Please specify a cluster with the --cluster flag or update current_cluster in the config",
+					"Please specify the context with the '--context' flag or set 'current_context' in the config",
-				cli.config.CurrentCluster,
+				cli.Config.Path(),
 			)
 		}
-		clusterName = cli.config.CurrentCluster
+		if _, ok := cli.Config.Contexts[cli.Config.CurrentContext]; !ok {
+			return nil, fmt.Errorf(
+				"current cluster context '%s' not found in the Uncloud config (%s). "+
+					"Please specify the context with the '--context' flag or update 'current_context' in the config",
+				cli.Config.CurrentContext,
+				cli.Config.Path(),
+			)
+		}
+		contextName = cli.Config.CurrentContext
 	}
 
-	cfg, ok := cli.config.Clusters[clusterName]
+	cfg, ok := cli.Config.Contexts[contextName]
 	if !ok {
-		return nil, fmt.Errorf("cluster %q not found in the config", clusterName)
+		return nil, fmt.Errorf("cluster context '%s' not found in the Uncloud config (%s)",
+			contextName, cli.Config.Path())
 	}
 	if len(cfg.Connections) == 0 {
-		return nil, fmt.Errorf("no connection configurations found for cluster %q in the config", clusterName)
+		return nil, fmt.Errorf(
+			"no connection configurations found for cluster context '%s' in the Uncloud config (%s)",
+			contextName, cli.Config.Path(),
+		)
 	}
 
 	// TODO: iterate over all connections and try to connect to the cluster using the first successful connection.
 	conn := cfg.Connections[0]
+
+	c, err := connectCluster(ctx, conn)
+	if err != nil {
+		return nil, fmt.Errorf("connect to cluster (context '%s'): %w", contextName, err)
+	}
+
+	return c, nil
+}
+
+func connectCluster(ctx context.Context, conn config.MachineConnection) (*client.Client, error) {
 	if conn.SSH != "" {
 		user, host, port, err := conn.SSH.Parse()
 		if err != nil {
@@ -103,45 +138,42 @@ func (cli *CLI) ConnectCluster(ctx context.Context, clusterName string) (*client
 			KeyPath: keyPath,
 		}
 		return client.New(ctx, connector.NewSSHConnector(sshConfig))
-	} else if conn.TCP.IsValid() {
+	} else if conn.TCP != nil && conn.TCP.IsValid() {
-		return client.New(ctx, connector.NewTCPConnector(conn.TCP))
+		return client.New(ctx, connector.NewTCPConnector(*conn.TCP))
 	}
-	return nil, errors.New("no valid connection configuration found for the cluster")
+
+	return nil, errors.New("connection configuration is invalid")
+}
+
+type InitClusterOptions struct {
+	Context       string
+	MachineName   string
+	Network       netip.Prefix
+	PublicIP      *netip.Addr
+	RemoteMachine *RemoteMachine
+	Version       string
 }
 
 // InitCluster initialises a new cluster on a remote machine and returns a client to interact with the cluster.
 // The client should be closed after use by the caller.
-func (cli *CLI) InitCluster(
+func (cli *CLI) InitCluster(ctx context.Context, opts InitClusterOptions) (*client.Client, error) {
-	ctx context.Context,
+	if opts.RemoteMachine != nil {
-	remoteMachine *RemoteMachine,
+		return cli.initRemoteMachine(ctx, opts)
-	clusterName,
-	machineName string,
-	netPrefix netip.Prefix,
-	publicIP *netip.Addr,
-) (*client.Client, error) {
-	if remoteMachine != nil {
-		return cli.initRemoteMachine(ctx, *remoteMachine, clusterName, machineName, netPrefix, publicIP)
 	}
 	// TODO: implement local machine initialisation
 	return nil, fmt.Errorf("local machine initialisation is not implemented yet")
 }
 
-func (cli *CLI) initRemoteMachine(
+func (cli *CLI) initRemoteMachine(ctx context.Context, opts InitClusterOptions) (*client.Client, error) {
-	ctx context.Context,
+	contextName := opts.Context
-	remoteMachine RemoteMachine,
+	if contextName == "" {
-	clusterName,
+		contextName = defaultContextName
-	machineName string,
-	netPrefix netip.Prefix,
-	publicIP *netip.Addr,
-) (*client.Client, error) {
-	if clusterName == "" {
-		clusterName = defaultClusterName
 	}
-	if _, ok := cli.config.Clusters[clusterName]; ok {
+	if _, ok := cli.Config.Contexts[contextName]; ok {
-		return nil, fmt.Errorf("cluster %q already exists", clusterName)
+		return nil, fmt.Errorf("cluster context '%s' already exists", contextName)
 	}
 
-	machineClient, err := cli.provisionRemoteMachine(ctx, remoteMachine)
+	machineClient, err := cli.provisionRemoteMachine(ctx, *opts.RemoteMachine, opts.Version)
 	if err != nil {
 		return nil, err
 	}
@@ -163,13 +195,24 @@ func (cli *CLI) initRemoteMachine(
 		}
 	}
 
-	req := &pb.InitClusterRequest{
+	// Check machine meets all necessary system requirements before proceeding.
-		MachineName: machineName,
+	checkResp, err := machineClient.CheckPrerequisites(ctx, &emptypb.Empty{})
-		Network:     pb.NewIPPrefix(netPrefix),
+	// TODO(lhf): remove Unimplemented check when v0.9.0 is released.
+	if err != nil {
+		if status.Convert(err).Code() != codes.Unimplemented {
+			return nil, fmt.Errorf("check machine prerequisites: %w", err)
+		}
+	} else if !checkResp.Satisfied {
+		return nil, fmt.Errorf("machine prerequisites not satisfied: %s", checkResp.Error)
 	}
-	if publicIP != nil {
+
-		if publicIP.IsValid() {
+	req := &pb.InitClusterRequest{
-			req.PublicIpConfig = &pb.InitClusterRequest_PublicIp{PublicIp: pb.NewIP(*publicIP)}
+		MachineName: opts.MachineName,
+		Network:     pb.NewIPPrefix(opts.Network),
+	}
+	if opts.PublicIP != nil {
+		if opts.PublicIP.IsValid() {
+			req.PublicIpConfig = &pb.InitClusterRequest_PublicIp{PublicIp: pb.NewIP(*opts.PublicIP)}
 		} else {
 			// Invalid or in other words zero IP means to automatically detect the public IP.
 			req.PublicIpConfig = &pb.InitClusterRequest_PublicIpAuto{PublicIpAuto: true}
@@ -180,44 +223,58 @@ func (cli *CLI) initRemoteMachine(
 	if err != nil {
 		return nil, fmt.Errorf("init cluster: %w", err)
 	}
-	fmt.Printf("Cluster %q initialised with machine %q\n", clusterName, resp.Machine.Name)
+	fmt.Printf("Cluster initialised with machine '%s' and saved as context '%s' in your local config (%s)\n",
-
+		resp.Machine.Name, contextName, cli.Config.Path())
-	if err = cli.CreateCluster(clusterName); err != nil {
+	if err = cli.CreateContext(contextName); err != nil {
-		return nil, fmt.Errorf("save cluster to config: %w", err)
+		return nil, fmt.Errorf("save cluster context to config: %w", err)
 	}
-	// Set the current cluster to the just created one if it is the only cluster in the config.
+	if err = cli.SetCurrentContext(contextName); err != nil {
-	if len(cli.config.Clusters) == 1 {
+		return nil, fmt.Errorf("set current cluster context: %w", err)
-		if err = cli.SetCurrentCluster(clusterName); err != nil {
-			return nil, fmt.Errorf("set current cluster: %w", err)
-		}
 	}
+	fmt.Printf("Current cluster context is now '%s'.\n", contextName)
 
-	// Save the machine's SSH connection details in the cluster config.
+	// Save the machine's SSH connection details in the context config.
 	connCfg := config.MachineConnection{
-		SSH:        config.NewSSHDestination(remoteMachine.User, remoteMachine.Host, remoteMachine.Port),
+		SSH:        config.NewSSHDestination(opts.RemoteMachine.User, opts.RemoteMachine.Host, opts.RemoteMachine.Port),
-		SSHKeyFile: remoteMachine.KeyPath,
+		SSHKeyFile: opts.RemoteMachine.KeyPath,
 	}
-	cli.config.Clusters[clusterName].Connections = append(cli.config.Clusters[clusterName].Connections, connCfg)
+	cli.Config.Contexts[contextName].Connections = append(cli.Config.Contexts[contextName].Connections, connCfg)
-	if err = cli.config.Save(); err != nil {
+	if err = cli.Config.Save(); err != nil {
 		return nil, fmt.Errorf("save config: %w", err)
 	}
 	return machineClient, nil
 }
 
-// AddMachine provisions a remote machine and adds it to the cluster. It returns a client to interact with the machine
+type AddMachineOptions struct {
-// which should be closed after use by the caller.
+	Context       string
-func (cli *CLI) AddMachine(
+	MachineName   string
-	ctx context.Context, remoteMachine RemoteMachine, clusterName, machineName string, publicIP *netip.Addr,
+	PublicIP      *netip.Addr
-) (*client.Client, error) {
+	RemoteMachine RemoteMachine
-	c, err := cli.ConnectCluster(ctx, clusterName)
+	Version       string
-	if err != nil {
+}
-		return nil, fmt.Errorf("connect to cluster: %w", err)
-	}
-	defer c.Close()
 
-	machineClient, err := cli.provisionRemoteMachine(ctx, remoteMachine)
+// AddMachine provisions a remote machine and adds it to the cluster. It returns a cluster client and a machine client.
+// The cluster client is connected to the existing machine in the cluster. It was used to add the new machine to the
+// cluster. The machine client is connected to the new machine and can be used to interact with it.
+// Both client should be closed after use by the caller.
+func (cli *CLI) AddMachine(ctx context.Context, opts AddMachineOptions) (*client.Client, *client.Client, error) {
+	contextName := opts.Context
+	if contextName == "" {
+		contextName = cli.Config.CurrentContext
+	}
+	c, err := cli.ConnectCluster(ctx, contextName)
 	if err != nil {
-		return nil, err
+		return nil, nil, fmt.Errorf("connect to cluster (context '%s'): %w", contextName, err)
+	}
+	defer func() {
+		if err != nil {
+			c.Close()
+		}
+	}()
+
+	machineClient, err := cli.provisionRemoteMachine(ctx, opts.RemoteMachine, opts.Version)
+	if err != nil {
+		return nil, nil, err
 	}
 	defer func() {
 		if err != nil {
@@ -228,21 +285,32 @@ func (cli *CLI) AddMachine(
 	// Check if the machine is already initialised as a cluster member and prompt the user to reset it first.
 	minfo, err := machineClient.Inspect(ctx, &emptypb.Empty{})
 	if err != nil {
-		return nil, fmt.Errorf("inspect machine: %w", err)
+		return nil, nil, fmt.Errorf("inspect machine: %w", err)
 	}
 	if minfo.Id != "" {
 		if err = cli.promptResetMachine(); err != nil {
-			return nil, err
+			return nil, nil, err
 		}
 	}
 
+	// Check machine meets all necessary system requirements before proceeding.
+	checkResp, err := machineClient.CheckPrerequisites(ctx, &emptypb.Empty{})
+	// TODO(lhf): remove Unimplemented check when v0.9.0 is released.
+	if err != nil {
+		if status.Convert(err).Code() != codes.Unimplemented {
+			return nil, nil, fmt.Errorf("check machine prerequisites: %w", err)
+		}
+	} else if !checkResp.Satisfied {
+		return nil, nil, fmt.Errorf("machine prerequisites not satisfied: %s", checkResp.Error)
+	}
+
 	tokenResp, err := machineClient.Token(ctx, &emptypb.Empty{})
 	if err != nil {
-		return nil, fmt.Errorf("get remote machine token: %w", err)
+		return nil, nil, fmt.Errorf("get remote machine token: %w", err)
 	}
 	token, err := machine.ParseToken(tokenResp.Token)
 	if err != nil {
-		return nil, fmt.Errorf("parse remote machine token: %w", err)
+		return nil, nil, fmt.Errorf("parse remote machine token: %w", err)
 	}
 
 	// Register the machine in the cluster using its public key and endpoints from the token.
@@ -251,15 +319,15 @@ func (cli *CLI) AddMachine(
 		endpoints[i] = pb.NewIPPort(addrPort)
 	}
 	addReq := &pb.AddMachineRequest{
-		Name: machineName,
+		Name: opts.MachineName,
 		Network: &pb.NetworkConfig{
 			Endpoints: endpoints,
 			PublicKey: token.PublicKey,
 		},
 	}
-	if publicIP != nil {
+	if opts.PublicIP != nil {
-		if publicIP.IsValid() {
+		if opts.PublicIP.IsValid() {
-			addReq.PublicIp = pb.NewIP(*publicIP)
+			addReq.PublicIp = pb.NewIP(*opts.PublicIP)
 		} else if token.PublicIP.IsValid() {
 			// Invalid or in other words zero IP means to use an automatically detected public IP from the token.
 			addReq.PublicIp = pb.NewIP(token.PublicIP)
@@ -268,13 +336,13 @@ func (cli *CLI) AddMachine(
 
 	addResp, err := c.AddMachine(ctx, addReq)
 	if err != nil {
-		return nil, fmt.Errorf("add machine to cluster: %w", err)
+		return nil, nil, fmt.Errorf("add machine to cluster (context '%s'): %w", contextName, err)
 	}
 
 	// List other machines in the cluster to include them in the join request.
-	machines, err := c.ListMachines(ctx)
+	machines, err := c.ListMachines(ctx, nil)
 	if err != nil {
-		return nil, fmt.Errorf("list cluster machines: %w", err)
+		return nil, nil, fmt.Errorf("list cluster machines: %w", err)
 	}
 	otherMachines := make([]*pb.MachineInfo, 0, len(machines)-1)
 	for _, m := range machines {
@@ -289,30 +357,34 @@ func (cli *CLI) AddMachine(
 		OtherMachines: otherMachines,
 	}
 	if _, err = machineClient.JoinCluster(ctx, joinReq); err != nil {
-		return nil, fmt.Errorf("join cluster: %w", err)
+		return nil, nil, fmt.Errorf("join cluster: %w", err)
 	}
 
-	fmt.Printf("Machine %q added to cluster\n", addResp.Machine.Name)
+	// TODO: fix empty context name when using the current context (contextName == "").
+	fmt.Printf("Machine '%s' added to the cluster (context '%s').\n", addResp.Machine.Name, contextName)
 
-	// Save the machine's SSH connection details in the cluster config.
+	// Save the machine's SSH connection details in the context config.
 	connCfg := config.MachineConnection{
-		SSH:        config.NewSSHDestination(remoteMachine.User, remoteMachine.Host, remoteMachine.Port),
+		SSH:        config.NewSSHDestination(opts.RemoteMachine.User, opts.RemoteMachine.Host, opts.RemoteMachine.Port),
-		SSHKeyFile: remoteMachine.KeyPath,
+		SSHKeyFile: opts.RemoteMachine.KeyPath,
 	}
-	if clusterName == "" {
+	if contextName == "" {
-		clusterName = cli.config.CurrentCluster
+		contextName = cli.Config.CurrentContext
 	}
-	cli.config.Clusters[clusterName].Connections = append(cli.config.Clusters[clusterName].Connections, connCfg)
+	cli.Config.Contexts[contextName].Connections = append(cli.Config.Contexts[contextName].Connections, connCfg)
-	if err = cli.config.Save(); err != nil {
+	if err = cli.Config.Save(); err != nil {
-		return nil, fmt.Errorf("save config: %w", err)
+		return nil, nil, fmt.Errorf("save config: %w", err)
 	}
 
-	return machineClient, nil
+	return c, machineClient, nil
 }
 
 // provisionRemoteMachine installs the Uncloud daemon and dependencies on the remote machine over SSH and returns
 // a machine API client to interact with the machine. The client should be closed after use by the caller.
-func (cli *CLI) provisionRemoteMachine(ctx context.Context, remoteMachine RemoteMachine) (*client.Client, error) {
+// The version parameter specifies the version of the Uncloud daemon to install. If empty, the latest version is used.
+func (cli *CLI) provisionRemoteMachine(
+	ctx context.Context, remoteMachine RemoteMachine, version string,
+) (*client.Client, error) {
 	// Provision the remote machine by installing the Uncloud daemon and dependencies over SSH.
 	sshClient, err := sshexec.Connect(remoteMachine.User, remoteMachine.Host, remoteMachine.Port, remoteMachine.KeyPath)
 	if err != nil {
@@ -323,7 +395,7 @@ func (cli *CLI) provisionRemoteMachine(ctx context.Context, remoteMachine Remote
 	}
 	exec := sshexec.NewRemote(sshClient)
 	// Install and run the Uncloud daemon and dependencies on the remote machine.
-	if err = provisionMachine(ctx, exec); err != nil {
+	if err = provisionMachine(ctx, exec, version); err != nil {
 		return nil, fmt.Errorf("provision machine: %w", err)
 	}
 
@@ -360,7 +432,7 @@ func (cli *CLI) promptResetMachine() error {
 				Negative("No").
 				Value(&confirm),
 		),
-	)
+	).WithAccessible(true)
 	if err := form.Run(); err != nil {
 		return fmt.Errorf("prompt user to confirm: %w", err)
 	}
@@ -369,7 +441,8 @@ func (cli *CLI) promptResetMachine() error {
 		return fmt.Errorf("remote machine is already initialised as a cluster member")
 	}
 	// TODO: implement resetting the remote machine.
-	return fmt.Errorf("resetting the remote machine is not implemented yet")
+	return fmt.Errorf("resetting the remote machine is not implemented yet. " +
+		"Please manually run 'uncloud-uninstall' on the remote machine to fully uninstall Uncloud from it")
 }
 
 // ProgressOut returns an output stream for progress writer.

internal/cli/client/caddy_test.go

@@ -1,69 +0,0 @@
-package client
-
-import (
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-	"testing"
-	"uncloud/internal/api"
-)
-
-func TestClient_NewCaddyDeployment(t *testing.T) {
-	t.Parallel()
-
-	cli := &Client{}
-
-	t.Run("latest image from Docker Hub", func(t *testing.T) {
-		t.Parallel()
-
-		deploy, err := cli.NewCaddyDeployment("", nil)
-		require.NoError(t, err)
-
-		assert.Equal(t, "caddy", deploy.Spec.Name)
-		assert.Equal(t, api.ServiceModeGlobal, deploy.Spec.Mode)
-		assert.Regexp(t, `^caddy:2\.\d+\.\d+$`, deploy.Spec.Container.Image)
-		expectedPorts := []api.PortSpec{
-			{
-				PublishedPort: 80,
-				ContainerPort: 80,
-				Protocol:      api.ProtocolTCP,
-				Mode:          api.PortModeHost,
-			},
-			{
-				PublishedPort: 443,
-				ContainerPort: 443,
-				Protocol:      api.ProtocolTCP,
-				Mode:          api.PortModeHost,
-			},
-		}
-		assert.Equal(t, expectedPorts, deploy.Spec.Ports)
-		// TODO:
-		//assert.Equal(t, alwaysPullImage, deploy.Spec.Container.PullPolicy)
-	})
-
-	t.Run("custom image", func(t *testing.T) {
-		t.Parallel()
-
-		image := "my-caddy:1.2.3"
-		deploy, err := cli.NewCaddyDeployment(image, nil)
-		require.NoError(t, err)
-
-		assert.Equal(t, "caddy", deploy.Spec.Name)
-		assert.Equal(t, api.ServiceModeGlobal, deploy.Spec.Mode)
-		assert.Equal(t, image, deploy.Spec.Container.Image)
-		expectedPorts := []api.PortSpec{
-			{
-				PublishedPort: 80,
-				ContainerPort: 80,
-				Protocol:      api.ProtocolTCP,
-				Mode:          api.PortModeHost,
-			},
-			{
-				PublishedPort: 443,
-				ContainerPort: 443,
-				Protocol:      api.ProtocolTCP,
-				Mode:          api.PortModeHost,
-			},
-		}
-		assert.Equal(t, expectedPorts, deploy.Spec.Ports)
-	})
-}

internal/cli/client/cluster.go

@@ -1,30 +0,0 @@
-package client
-
-import (
-	"context"
-	"google.golang.org/protobuf/types/known/emptypb"
-	"uncloud/internal/machine/api/pb"
-)
-
-func (cli *Client) InspectMachine(ctx context.Context, id string) (*pb.MachineMember, error) {
-	machines, err := cli.ListMachines(ctx)
-	if err != nil {
-		return nil, err
-	}
-
-	for _, m := range machines {
-		if m.Machine.Id == id || m.Machine.Name == id {
-			return m, nil
-		}
-	}
-
-	return nil, ErrNotFound
-}
-
-func (cli *Client) ListMachines(ctx context.Context) ([]*pb.MachineMember, error) {
-	resp, err := cli.ClusterClient.ListMachines(ctx, &emptypb.Empty{})
-	if err != nil {
-		return nil, err
-	}
-	return resp.Machines, nil
-}

internal/cli/client/resolver.go

@@ -1,116 +0,0 @@
-package client
-
-import (
-	"fmt"
-	"github.com/distribution/reference"
-	"strings"
-	"uncloud/internal/api"
-	"uncloud/internal/secret"
-)
-
-// ServiceSpecResolver transforms user-provided service specs into deployment-ready form.
-type ServiceSpecResolver struct {
-	ClusterDomain string
-}
-
-func NewServiceSpecResolver(clusterDomain string) *ServiceSpecResolver {
-	return &ServiceSpecResolver{ClusterDomain: clusterDomain}
-}
-
-// Resolve transforms a service spec into its fully resolved form ready for deployment.
-func (r *ServiceSpecResolver) Resolve(spec *api.ServiceSpec) error {
-	if err := spec.Validate(); err != nil {
-		return fmt.Errorf("invalid service spec: %w", err)
-	}
-
-	steps := []func(*api.ServiceSpec) error{
-		r.applyDefaults,
-		r.resolveServiceName,
-		r.expandIngressPorts,
-	}
-
-	for _, step := range steps {
-		if err := step(spec); err != nil {
-			return err
-		}
-	}
-
-	return nil
-}
-
-func (r *ServiceSpecResolver) applyDefaults(spec *api.ServiceSpec) error {
-	if spec.Mode == "" {
-		spec.Mode = api.ServiceModeReplicated
-	}
-	// Ensure the replicated service has at least one replica.
-	if spec.Mode == api.ServiceModeReplicated && spec.Replicas == 0 {
-		spec.Replicas = 1
-	}
-
-	return nil
-}
-
-func (r *ServiceSpecResolver) resolveServiceName(spec *api.ServiceSpec) error {
-	if spec.Name != "" {
-		return nil
-	}
-
-	// Generate a random service name from the image when not provided.
-	img, err := reference.ParseDockerRef(spec.Container.Image)
-	if err != nil {
-		return fmt.Errorf("invalid image: %w", err)
-	}
-	// Get the image name without the repository and tag/digest parts.
-	imageName := reference.FamiliarName(img)
-	// Get the last part of the image name (path), e.g. "nginx" from "bitnami/nginx".
-	if i := strings.LastIndex(imageName, "/"); i != -1 {
-		imageName = imageName[i+1:]
-	}
-	// Append a random suffix to the image name to generate an optimistically unique service name.
-	suffix, err := secret.RandomAlphaNumeric(4)
-	if err != nil {
-		return fmt.Errorf("generate random suffix: %w", err)
-	}
-	spec.Name = fmt.Sprintf("%s-%s", imageName, suffix)
-
-	return nil
-}
-
-// expandIngressPorts processes HTTP(S) ingress ports in a service spec by:
-// 1. Setting a default hostname (service-name.cluster-domain) for ports without a hostname.
-// 2. Duplicating a port with a cluster domain hostname for ports with external domains.
-// This ensures every ingress port is accessible via the cluster domain, while preserving any custom domains specified
-// by the user.
-func (r *ServiceSpecResolver) expandIngressPorts(spec *api.ServiceSpec) error {
-	for i, port := range spec.Ports {
-		if port.Protocol != api.ProtocolHTTP && port.Protocol != api.ProtocolHTTPS {
-			continue
-		}
-
-		if port.Hostname == "" {
-			if r.ClusterDomain == "" {
-				return fmt.Errorf("cluster domain must be reserved to generate hostname for ingress port: %d/%s",
-					port.ContainerPort, port.Protocol)
-			}
-			// Assign the default hostname (service-name.cluster-domain).
-			spec.Ports[i].Hostname = fmt.Sprintf("%s.%s", spec.Name, r.ClusterDomain)
-		} else {
-			if r.ClusterDomain == "" {
-				// When no cluster domain is reserved, use only the provided hostname.
-				continue
-			}
-
-			if strings.HasSuffix(port.Hostname, "."+r.ClusterDomain) {
-				// If the hostname is already a cluster subdomain, use as is.
-				continue
-			}
-			// For external domains, duplicate the port with a service-name.cluster-domain hostname so the service
-			// can be accessed via both hostnames.
-			newPort := port
-			newPort.Hostname = fmt.Sprintf("%s.%s", spec.Name, r.ClusterDomain)
-			spec.Ports = append(spec.Ports, newPort)
-		}
-	}
-
-	return nil
-}

internal/cli/config/cluster.go

@@ -1,6 +0,0 @@
-package config
-
-type Cluster struct {
-	Name        string              `toml:"-"`
-	Connections []MachineConnection `toml:"connections"`
-}

internal/cli/config/config.go

@@ -2,14 +2,15 @@ package config
 
 import (
 	"fmt"
-	"github.com/BurntSushi/toml"
 	"os"
 	"path/filepath"
+
+	"github.com/goccy/go-yaml"
 )
 
 type Config struct {
-	Clusters       map[string]*Cluster `toml:"clusters"`
+	CurrentContext string              `yaml:"current_context"`
-	CurrentCluster string              `toml:"current_cluster"`
+	Contexts       map[string]*Context `yaml:"contexts"`
 
 	// path is the file path config is read from.
 	path string
@@ -18,10 +19,10 @@ type Config struct {
 func NewFromFile(path string) (*Config, error) {
 	_, err := os.Stat(path)
 	if err != nil && !os.IsNotExist(err) {
-		return nil, fmt.Errorf("check file permissions %q: %w", path, err)
+		return nil, fmt.Errorf("check file permissions '%s': %w", path, err)
 	}
 	c := &Config{
-		Clusters: map[string]*Cluster{},
+		Contexts: map[string]*Context{},
 		path:     path,
 	}
 	if os.IsNotExist(err) {
@@ -34,30 +35,37 @@ func NewFromFile(path string) (*Config, error) {
 	return c, nil
 }
 
+func (c *Config) Path() string {
+	return c.path
+}
+
 func (c *Config) Read() error {
-	_, err := toml.DecodeFile(c.path, c)
+	data, err := os.ReadFile(c.path)
 	if err != nil {
-		return fmt.Errorf("read config file %q: %w", c.path, err)
+		return fmt.Errorf("read config file '%s': %w", c.path, err)
 	}
+	if err = yaml.Unmarshal(data, c); err != nil {
+		return fmt.Errorf("parse config file '%s': %s", c.path, yaml.FormatError(err, true, true))
+	}
+
 	return nil
 }
 
 func (c *Config) Save() error {
 	dir, _ := filepath.Split(c.path)
 	if err := os.MkdirAll(dir, 0700); err != nil {
-		return fmt.Errorf("create config directory %q: %w", dir, err)
+		return fmt.Errorf("create config directory '%s': %w", dir, err)
 	}
 
 	f, err := os.OpenFile(c.path, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0600)
 	if err != nil {
-		return fmt.Errorf("write config file %q: %w", c.path, err)
+		return fmt.Errorf("write config file '%s': %w", c.path, err)
 	}
 
-	encoder := toml.NewEncoder(f)
+	encoder := yaml.NewEncoder(f, yaml.Indent(2), yaml.IndentSequence(true))
-	encoder.Indent = ""
 	if err = encoder.Encode(c); err != nil {
 		_ = f.Close()
-		return fmt.Errorf("encode config file %q: %w", c.path, err)
+		return fmt.Errorf("encode config file '%s': %w", c.path, err)
 	}
 	return f.Close()
 }

internal/cli/config/connection.go

@@ -5,7 +5,8 @@ import (
 	"net/netip"
 	"strconv"
 	"strings"
-	"uncloud/internal/secret"
+
+	"github.com/psviderski/uncloud/internal/secret"
 )
 
 const (
@@ -14,11 +15,13 @@ const (
 )
 
 type MachineConnection struct {
-	SSH        SSHDestination `toml:"ssh,omitempty"`
+	SSH        SSHDestination `yaml:"ssh,omitempty"`
-	TCP        netip.AddrPort `toml:"tcp,omitempty"`
+	SSHKeyFile string         `yaml:"ssh_key_file,omitempty"`
-	Host       string         `toml:"host,omitempty"`
+	// TCP is the address and port of the machine's API server.
-	PublicKey  secret.Secret  `toml:"public_key,omitempty"`
+	// The pointer is used to omit the field when not set. Otherwise, yaml marshalling includes an empty object.
-	SSHKeyFile string         `toml:"ssh_key_file,omitempty"`
+	TCP       *netip.AddrPort `yaml:"tcp,omitempty"`
+	Host      string          `yaml:"host,omitempty"`
+	PublicKey secret.Secret   `yaml:"public_key,omitempty"`
 }
 
 // SSHDestination represents an SSH destination string in the canonical form of "user@host:port".

internal/cli/config/context.go

@@ -0,0 +1,6 @@
+package config
+
+type Context struct {
+	Name        string              `yaml:"-"`
+	Connections []MachineConnection `yaml:"connections"`
+}

internal/cli/flags.go

@@ -0,0 +1,23 @@
+package cli
+
+import (
+	"strings"
+)
+
+// ExpandCommaSeparatedValues takes a slice of strings and expands any comma-separated values into individual elements.
+func ExpandCommaSeparatedValues(values []string) []string {
+	if len(values) == 0 {
+		return nil
+	}
+
+	var expanded []string
+	for _, value := range values {
+		for _, v := range strings.Split(value, ",") {
+			if v = strings.TrimSpace(v); v != "" {
+				expanded = append(expanded, v)
+			}
+		}
+	}
+
+	return expanded
+}

internal/cli/machine.go

@@ -4,7 +4,9 @@ import (
 	"context"
 	"fmt"
 	"os"
-	"uncloud/internal/sshexec"
+	"strings"
+
+	"github.com/psviderski/uncloud/internal/sshexec"
 )
 
 // TODO: support pinning the script version to the CLI version.
@@ -17,36 +19,38 @@ type RemoteMachine struct {
 	KeyPath string
 }
 
-func installCmd(user string) string {
+func installCmd(user string, version string) string {
-	sudoPrefix := "sudo"
+	sudoPrefix := ""
+	var env []string
+
 	// Add the SSH user (non-root) to the uncloud group to allow access to the Uncloud daemon unix socket.
-	env := "UNCLOUD_GROUP_ADD_USER=" + user
+	if user != "root" {
-
+		sudoPrefix = "sudo"
-	curlBashCmd := fmt.Sprintf(
+		env = append(env, "UNCLOUD_GROUP_ADD_USER="+sshexec.Quote(user))
-		"curl -fsSL %s | %s %s bash", sshexec.Quote(installScriptURL), sudoPrefix, sshexec.Quote(env),
-	)
-
-	if user == "root" {
-		curlBashCmd = fmt.Sprintf(
-			"curl -fsSL %s | bash", sshexec.Quote(installScriptURL),
-		)
 	}
+	if version != "" {
+		env = append(env, "UNCLOUD_VERSION="+sshexec.Quote(version))
+	}
+
+	envCmd := strings.Join(env, " ")
+	curlBashCmd := fmt.Sprintf("curl -fsSL %s | %s %s bash", sshexec.Quote(installScriptURL), sudoPrefix, envCmd)
 
 	return curlBashCmd
 }
 
 // provisionMachine provisions the remote machine by downloading the Uncloud install script from GitHub and running it.
-func provisionMachine(ctx context.Context, exec sshexec.Executor) error {
+// If version is specified, it will be passed to the install script as UNCLOUD_VERSION environment variable.
+func provisionMachine(ctx context.Context, exec sshexec.Executor, version string) error {
 	user, err := exec.Run(ctx, "whoami")
 	if err != nil {
 		return fmt.Errorf("run whoami: %w", err)
 	}
 
-	installCmd := installCmd(user)
+	cmd := installCmd(user, version)
 
 	fmt.Println("Downloading Uncloud install script:", installScriptURL)
 
-	cmd := sshexec.QuoteCommand("bash", "-c", "set -o pipefail; "+installCmd)
+	cmd = sshexec.QuoteCommand("bash", "-c", "set -o pipefail; "+cmd)
 	if err = exec.Stream(ctx, cmd, os.Stdout, os.Stderr); err != nil {
 		return fmt.Errorf("download and run install script: %w", err)
 	}

internal/cli/machine_test.go

@@ -8,12 +8,29 @@ import (
 
 func TestInstallCmd(t *testing.T) {
 	t.Run("root", func(t *testing.T) {
-		cmd := installCmd("root")
+		cmd := installCmd("root", "")
 		assert.NotContains(t, cmd, "sudo")
+		assert.NotContains(t, cmd, "UNCLOUD_GROUP_ADD_USER")
+	})
+
+	// Test with version
+	t.Run("root with version", func(t *testing.T) {
+		cmd := installCmd("root", "v1.2.3")
+		assert.NotContains(t, cmd, "sudo")
+		assert.NotContains(t, cmd, "UNCLOUD_GROUP_ADD_USER")
+		assert.Contains(t, cmd, "UNCLOUD_VERSION=v1.2.3")
 	})
 
 	t.Run("nonroot", func(t *testing.T) {
-		cmd := installCmd("nonroot")
+		cmd := installCmd("nonroot", "")
 		assert.Contains(t, cmd, "sudo")
+		assert.Contains(t, cmd, "UNCLOUD_GROUP_ADD_USER=nonroot")
+	})
+
+	t.Run("nonroot with version", func(t *testing.T) {
+		cmd := installCmd("nonroot", "v1.2.3")
+		assert.Contains(t, cmd, "sudo")
+		assert.Contains(t, cmd, "UNCLOUD_GROUP_ADD_USER=nonroot")
+		assert.Contains(t, cmd, "UNCLOUD_VERSION=v1.2.3")
 	})
 }

internal/cli/prompt.go

@@ -0,0 +1,23 @@
+package cli
+
+import "github.com/charmbracelet/huh"
+
+func Confirm() (bool, error) {
+	var confirmed bool
+	form := huh.NewForm(
+		huh.NewGroup(
+			huh.NewConfirm().
+				Title(
+					"Do you want to continue?",
+				).
+				Affirmative("Yes!").
+				Negative("No").
+				Value(&confirmed),
+		),
+	).WithAccessible(true)
+	if err := form.Run(); err != nil {
+		return false, err
+	}
+
+	return confirmed, nil
+}

internal/daemon/daemon.go

@@ -5,7 +5,7 @@ import (
 	"fmt"
 	systemd "github.com/coreos/go-systemd/daemon"
 	"log/slog"
-	"uncloud/internal/machine"
+	"github.com/psviderski/uncloud/internal/machine"
 )
 
 type Daemon struct {

internal/daemon/token.go

@@ -3,10 +3,10 @@ package daemon
 import (
 	"errors"
 	"fmt"
+	"github.com/psviderski/uncloud/internal/machine"
+	"github.com/psviderski/uncloud/internal/machine/network"
 	"net/netip"
 	"os"
-	"uncloud/internal/machine"
-	"uncloud/internal/machine/network"
 )
 
 // MachineToken returns the local machine's token that can be used for adding the machine to a cluster.

internal/log/handler.go

@@ -45,11 +45,19 @@ func (h *SlogTextHandler) Enabled(ctx context.Context, level slog.Level) bool {
 }
 
 func (h *SlogTextHandler) WithAttrs(attrs []slog.Attr) slog.Handler {
-	return h.TextHandler.WithAttrs(attrs)
+	return &SlogTextHandler{
+		TextHandler: h.TextHandler.WithAttrs(attrs).(*slog.TextHandler),
+		opts:        h.opts,
+		w:           h.w,
+	}
 }
 
 func (h *SlogTextHandler) WithGroup(name string) slog.Handler {
-	return h.TextHandler.WithGroup(name)
+	return &SlogTextHandler{
+		TextHandler: h.TextHandler.WithGroup(name).(*slog.TextHandler),
+		opts:        h.opts,
+		w:           h.w,
+	}
 }
 
 func (h *SlogTextHandler) Handle(ctx context.Context, r slog.Record) error {

internal/machine/api/pb/docker.proto

@@ -14,23 +14,37 @@ service Docker {
   rpc StopContainer(StopContainerRequest) returns (google.protobuf.Empty);
   rpc ListContainers(ListContainersRequest) returns (ListContainersResponse);
   rpc RemoveContainer(RemoveContainerRequest) returns (google.protobuf.Empty);
+
   rpc PullImage(PullImageRequest) returns (stream JSONMessage);
+  rpc InspectImage(InspectImageRequest) returns (InspectImageResponse);
+  // InspectRemoteImage returns the image metadata for an image in a remote registry using the machine's
+  // Docker auth credentials if necessary.
+  rpc InspectRemoteImage(InspectRemoteImageRequest) returns (InspectRemoteImageResponse);
+
+  rpc CreateVolume(CreateVolumeRequest) returns (CreateVolumeResponse);
+  rpc ListVolumes(ListVolumesRequest) returns (ListVolumesResponse);
+  rpc RemoveVolume(RemoveVolumeRequest) returns (google.protobuf.Empty);
+
+  rpc CreateServiceContainer(CreateServiceContainerRequest) returns (CreateContainerResponse);
+  rpc InspectServiceContainer(InspectContainerRequest) returns (ServiceContainer);
+  rpc ListServiceContainers(ListServiceContainersRequest) returns (ListServiceContainersResponse);
+  rpc RemoveServiceContainer(RemoveContainerRequest) returns (google.protobuf.Empty);
 }
 
 message CreateContainerRequest {
-  // JSON serialized container.Config.
+  // JSON serialised container.Config.
   bytes config = 1;
-  // JSON serialized container.HostConfig.
+  // JSON serialised container.HostConfig.
   bytes host_config = 2;
-  // JSON serialized network.NetworkingConfig.
+  // JSON serialised network.NetworkingConfig.
   bytes network_config = 3;
-  // JSON serialized ocispec.Platform.
+  // JSON serialised ocispec.Platform.
   bytes platform = 4;
   string name = 5;
 }
 
 message CreateContainerResponse {
-  // JSON serialized container.CreateResponse.
+  // JSON serialised container.CreateResponse.
   bytes response = 1;
 }
 
@@ -39,51 +53,145 @@ message InspectContainerRequest {
 }
 
 message InspectContainerResponse {
-  // JSON serialized container.InspectResponse.
+  // JSON serialised container.InspectResponse.
   bytes response = 1;
 }
 
 message StartContainerRequest {
   string id = 1;
-  // JSON serialized container.StartOptions.
+  // JSON serialised container.StartOptions.
   bytes options = 2;
 }
 
 message StopContainerRequest {
   string id = 1;
-  // JSON serialized container.StopOptions.
+  // JSON serialised container.StopOptions.
   bytes options = 2;
 }
 
 message ListContainersRequest {
-  // JSON serialized container.ListOptions.
+  // JSON serialised container.ListOptions.
   bytes options = 1;
 }
 
-// ListContainersResponse structure allows broadcasting ListContainers requests to multiple machines.
 message ListContainersResponse {
+  // Must contain only one repeated messages field to allow broadcasting ListContainers requests to multiple machines.
   repeated MachineContainers messages = 1;
 }
 
 message MachineContainers {
   Metadata metadata = 1;
-  // JSON serialized []container.ContainerJSON.
+  // JSON serialised []container.InspectResponse.
   bytes containers = 2;
 }
 
 message RemoveContainerRequest {
   string id = 1;
-  // JSON serialized container.RemoveOptions.
+  // JSON serialised container.RemoveOptions.
   bytes options = 2;
 }
 
 message PullImageRequest {
   string image = 1;
-  // JSON serialized image.PullOptions.
+  // JSON serialised image.PullOptions.
   bytes options = 2;
 }
 
 message JSONMessage {
-  // JSON serialized jsonmessage.JSONMessage.
+  // JSON serialised jsonmessage.JSONMessage.
   bytes message = 1;
 }
+
+message InspectImageRequest {
+  string id = 1;
+}
+
+message InspectImageResponse {
+  // Must contain only one repeated messages field to allow broadcasting InspectImage requests to multiple machines.
+  repeated Image messages = 1;
+}
+
+message Image {
+  Metadata metadata = 1;
+  // JSON serialised types.ImageInspect.
+  bytes image = 2;
+}
+
+message InspectRemoteImageRequest {
+  string id = 1;
+}
+
+message InspectRemoteImageResponse {
+  // Must contain only one repeated messages field to allow broadcasting InspectRemoteImage requests to multiple machines.
+  repeated RemoteImage messages = 1;
+}
+
+message RemoteImage {
+  Metadata metadata = 1;
+  // Image reference in the canonical form with the digest.
+  string reference = 2;
+  // Raw JSON manifest from the registry.
+  bytes manifest = 3;
+}
+
+message CreateVolumeRequest {
+  // JSON serialised volume.CreateOptions.
+  bytes options = 1;
+}
+
+message CreateVolumeResponse {
+  // JSON serialised volume.Volume.
+  bytes volume = 1;
+}
+
+message ListVolumesRequest {
+  // JSON serialised volume.ListOptions.
+  bytes options = 1;
+}
+
+message ListVolumesResponse {
+  // Must contain only one repeated messages field to allow broadcasting ListVolumes requests to multiple machines.
+  repeated MachineVolumes messages = 1;
+}
+
+message MachineVolumes {
+  Metadata metadata = 1;
+  // JSON serialised volume.ListResponse.
+  bytes response = 2;
+}
+
+message RemoveVolumeRequest {
+  string id = 1;
+  bool force = 2;
+}
+
+message CreateServiceContainerRequest {
+  string service_id = 1;
+  // JSON serialised api.ServiceSpec.
+  bytes service_spec = 2;
+  string container_name = 3;
+}
+
+message ServiceContainer {
+  // JSON serialised container.InspectResponse.
+  bytes container = 1;
+  // JSON serialised api.ServiceSpec.
+  bytes service_spec = 2;
+}
+
+message ListServiceContainersRequest {
+  string service_id = 1;
+  // JSON serialised container.ListOptions.
+  bytes options = 2;
+}
+
+message ListServiceContainersResponse {
+  // Must contain only one repeated messages field to allow broadcasting ListServiceContainers requests
+  // to multiple machines.
+  repeated MachineServiceContainers messages = 1;
+}
+
+message MachineServiceContainers {
+  Metadata metadata = 1;
+  repeated ServiceContainer containers = 2;
+}

internal/machine/api/pb/docker_grpc.pb.go

@@ -20,13 +20,22 @@ import (
 const _ = grpc.SupportPackageIsVersion9
 
 const (
-	Docker_CreateContainer_FullMethodName  = "/api.Docker/CreateContainer"
+	Docker_CreateContainer_FullMethodName         = "/api.Docker/CreateContainer"
-	Docker_InspectContainer_FullMethodName = "/api.Docker/InspectContainer"
+	Docker_InspectContainer_FullMethodName        = "/api.Docker/InspectContainer"
-	Docker_StartContainer_FullMethodName   = "/api.Docker/StartContainer"
+	Docker_StartContainer_FullMethodName          = "/api.Docker/StartContainer"
-	Docker_StopContainer_FullMethodName    = "/api.Docker/StopContainer"
+	Docker_StopContainer_FullMethodName           = "/api.Docker/StopContainer"
-	Docker_ListContainers_FullMethodName   = "/api.Docker/ListContainers"
+	Docker_ListContainers_FullMethodName          = "/api.Docker/ListContainers"
-	Docker_RemoveContainer_FullMethodName  = "/api.Docker/RemoveContainer"
+	Docker_RemoveContainer_FullMethodName         = "/api.Docker/RemoveContainer"
-	Docker_PullImage_FullMethodName        = "/api.Docker/PullImage"
+	Docker_PullImage_FullMethodName               = "/api.Docker/PullImage"
+	Docker_InspectImage_FullMethodName            = "/api.Docker/InspectImage"
+	Docker_InspectRemoteImage_FullMethodName      = "/api.Docker/InspectRemoteImage"
+	Docker_CreateVolume_FullMethodName            = "/api.Docker/CreateVolume"
+	Docker_ListVolumes_FullMethodName             = "/api.Docker/ListVolumes"
+	Docker_RemoveVolume_FullMethodName            = "/api.Docker/RemoveVolume"
+	Docker_CreateServiceContainer_FullMethodName  = "/api.Docker/CreateServiceContainer"
+	Docker_InspectServiceContainer_FullMethodName = "/api.Docker/InspectServiceContainer"
+	Docker_ListServiceContainers_FullMethodName   = "/api.Docker/ListServiceContainers"
+	Docker_RemoveServiceContainer_FullMethodName  = "/api.Docker/RemoveServiceContainer"
 )
 
 // DockerClient is the client API for Docker service.
@@ -40,6 +49,17 @@ type DockerClient interface {
 	ListContainers(ctx context.Context, in *ListContainersRequest, opts ...grpc.CallOption) (*ListContainersResponse, error)
 	RemoveContainer(ctx context.Context, in *RemoveContainerRequest, opts ...grpc.CallOption) (*emptypb.Empty, error)
 	PullImage(ctx context.Context, in *PullImageRequest, opts ...grpc.CallOption) (grpc.ServerStreamingClient[JSONMessage], error)
+	InspectImage(ctx context.Context, in *InspectImageRequest, opts ...grpc.CallOption) (*InspectImageResponse, error)
+	// InspectRemoteImage returns the image metadata for an image in a remote registry using the machine's
+	// Docker auth credentials if necessary.
+	InspectRemoteImage(ctx context.Context, in *InspectRemoteImageRequest, opts ...grpc.CallOption) (*InspectRemoteImageResponse, error)
+	CreateVolume(ctx context.Context, in *CreateVolumeRequest, opts ...grpc.CallOption) (*CreateVolumeResponse, error)
+	ListVolumes(ctx context.Context, in *ListVolumesRequest, opts ...grpc.CallOption) (*ListVolumesResponse, error)
+	RemoveVolume(ctx context.Context, in *RemoveVolumeRequest, opts ...grpc.CallOption) (*emptypb.Empty, error)
+	CreateServiceContainer(ctx context.Context, in *CreateServiceContainerRequest, opts ...grpc.CallOption) (*CreateContainerResponse, error)
+	InspectServiceContainer(ctx context.Context, in *InspectContainerRequest, opts ...grpc.CallOption) (*ServiceContainer, error)
+	ListServiceContainers(ctx context.Context, in *ListServiceContainersRequest, opts ...grpc.CallOption) (*ListServiceContainersResponse, error)
+	RemoveServiceContainer(ctx context.Context, in *RemoveContainerRequest, opts ...grpc.CallOption) (*emptypb.Empty, error)
 }
 
 type dockerClient struct {
@@ -129,6 +149,96 @@ func (c *dockerClient) PullImage(ctx context.Context, in *PullImageRequest, opts
 // This type alias is provided for backwards compatibility with existing code that references the prior non-generic stream type by name.
 type Docker_PullImageClient = grpc.ServerStreamingClient[JSONMessage]
 
+func (c *dockerClient) InspectImage(ctx context.Context, in *InspectImageRequest, opts ...grpc.CallOption) (*InspectImageResponse, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(InspectImageResponse)
+	err := c.cc.Invoke(ctx, Docker_InspectImage_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) InspectRemoteImage(ctx context.Context, in *InspectRemoteImageRequest, opts ...grpc.CallOption) (*InspectRemoteImageResponse, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(InspectRemoteImageResponse)
+	err := c.cc.Invoke(ctx, Docker_InspectRemoteImage_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) CreateVolume(ctx context.Context, in *CreateVolumeRequest, opts ...grpc.CallOption) (*CreateVolumeResponse, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(CreateVolumeResponse)
+	err := c.cc.Invoke(ctx, Docker_CreateVolume_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) ListVolumes(ctx context.Context, in *ListVolumesRequest, opts ...grpc.CallOption) (*ListVolumesResponse, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(ListVolumesResponse)
+	err := c.cc.Invoke(ctx, Docker_ListVolumes_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) RemoveVolume(ctx context.Context, in *RemoveVolumeRequest, opts ...grpc.CallOption) (*emptypb.Empty, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(emptypb.Empty)
+	err := c.cc.Invoke(ctx, Docker_RemoveVolume_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) CreateServiceContainer(ctx context.Context, in *CreateServiceContainerRequest, opts ...grpc.CallOption) (*CreateContainerResponse, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(CreateContainerResponse)
+	err := c.cc.Invoke(ctx, Docker_CreateServiceContainer_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) InspectServiceContainer(ctx context.Context, in *InspectContainerRequest, opts ...grpc.CallOption) (*ServiceContainer, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(ServiceContainer)
+	err := c.cc.Invoke(ctx, Docker_InspectServiceContainer_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) ListServiceContainers(ctx context.Context, in *ListServiceContainersRequest, opts ...grpc.CallOption) (*ListServiceContainersResponse, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(ListServiceContainersResponse)
+	err := c.cc.Invoke(ctx, Docker_ListServiceContainers_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *dockerClient) RemoveServiceContainer(ctx context.Context, in *RemoveContainerRequest, opts ...grpc.CallOption) (*emptypb.Empty, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(emptypb.Empty)
+	err := c.cc.Invoke(ctx, Docker_RemoveServiceContainer_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
 // DockerServer is the server API for Docker service.
 // All implementations must embed UnimplementedDockerServer
 // for forward compatibility.
@@ -140,6 +250,17 @@ type DockerServer interface {
 	ListContainers(context.Context, *ListContainersRequest) (*ListContainersResponse, error)
 	RemoveContainer(context.Context, *RemoveContainerRequest) (*emptypb.Empty, error)
 	PullImage(*PullImageRequest, grpc.ServerStreamingServer[JSONMessage]) error
+	InspectImage(context.Context, *InspectImageRequest) (*InspectImageResponse, error)
+	// InspectRemoteImage returns the image metadata for an image in a remote registry using the machine's
+	// Docker auth credentials if necessary.
+	InspectRemoteImage(context.Context, *InspectRemoteImageRequest) (*InspectRemoteImageResponse, error)
+	CreateVolume(context.Context, *CreateVolumeRequest) (*CreateVolumeResponse, error)
+	ListVolumes(context.Context, *ListVolumesRequest) (*ListVolumesResponse, error)
+	RemoveVolume(context.Context, *RemoveVolumeRequest) (*emptypb.Empty, error)
+	CreateServiceContainer(context.Context, *CreateServiceContainerRequest) (*CreateContainerResponse, error)
+	InspectServiceContainer(context.Context, *InspectContainerRequest) (*ServiceContainer, error)
+	ListServiceContainers(context.Context, *ListServiceContainersRequest) (*ListServiceContainersResponse, error)
+	RemoveServiceContainer(context.Context, *RemoveContainerRequest) (*emptypb.Empty, error)
 	mustEmbedUnimplementedDockerServer()
 }
 
@@ -171,6 +292,33 @@ func (UnimplementedDockerServer) RemoveContainer(context.Context, *RemoveContain
 func (UnimplementedDockerServer) PullImage(*PullImageRequest, grpc.ServerStreamingServer[JSONMessage]) error {
 	return status.Errorf(codes.Unimplemented, "method PullImage not implemented")
 }
+func (UnimplementedDockerServer) InspectImage(context.Context, *InspectImageRequest) (*InspectImageResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method InspectImage not implemented")
+}
+func (UnimplementedDockerServer) InspectRemoteImage(context.Context, *InspectRemoteImageRequest) (*InspectRemoteImageResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method InspectRemoteImage not implemented")
+}
+func (UnimplementedDockerServer) CreateVolume(context.Context, *CreateVolumeRequest) (*CreateVolumeResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method CreateVolume not implemented")
+}
+func (UnimplementedDockerServer) ListVolumes(context.Context, *ListVolumesRequest) (*ListVolumesResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method ListVolumes not implemented")
+}
+func (UnimplementedDockerServer) RemoveVolume(context.Context, *RemoveVolumeRequest) (*emptypb.Empty, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method RemoveVolume not implemented")
+}
+func (UnimplementedDockerServer) CreateServiceContainer(context.Context, *CreateServiceContainerRequest) (*CreateContainerResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method CreateServiceContainer not implemented")
+}
+func (UnimplementedDockerServer) InspectServiceContainer(context.Context, *InspectContainerRequest) (*ServiceContainer, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method InspectServiceContainer not implemented")
+}
+func (UnimplementedDockerServer) ListServiceContainers(context.Context, *ListServiceContainersRequest) (*ListServiceContainersResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method ListServiceContainers not implemented")
+}
+func (UnimplementedDockerServer) RemoveServiceContainer(context.Context, *RemoveContainerRequest) (*emptypb.Empty, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method RemoveServiceContainer not implemented")
+}
 func (UnimplementedDockerServer) mustEmbedUnimplementedDockerServer() {}
 func (UnimplementedDockerServer) testEmbeddedByValue()                {}
 
@@ -311,6 +459,168 @@ func _Docker_PullImage_Handler(srv interface{}, stream grpc.ServerStream) error
 // This type alias is provided for backwards compatibility with existing code that references the prior non-generic stream type by name.
 type Docker_PullImageServer = grpc.ServerStreamingServer[JSONMessage]
 
+func _Docker_InspectImage_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(InspectImageRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).InspectImage(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_InspectImage_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).InspectImage(ctx, req.(*InspectImageRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_InspectRemoteImage_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(InspectRemoteImageRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).InspectRemoteImage(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_InspectRemoteImage_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).InspectRemoteImage(ctx, req.(*InspectRemoteImageRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_CreateVolume_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(CreateVolumeRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).CreateVolume(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_CreateVolume_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).CreateVolume(ctx, req.(*CreateVolumeRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_ListVolumes_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(ListVolumesRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).ListVolumes(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_ListVolumes_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).ListVolumes(ctx, req.(*ListVolumesRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_RemoveVolume_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(RemoveVolumeRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).RemoveVolume(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_RemoveVolume_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).RemoveVolume(ctx, req.(*RemoveVolumeRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_CreateServiceContainer_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(CreateServiceContainerRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).CreateServiceContainer(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_CreateServiceContainer_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).CreateServiceContainer(ctx, req.(*CreateServiceContainerRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_InspectServiceContainer_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(InspectContainerRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).InspectServiceContainer(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_InspectServiceContainer_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).InspectServiceContainer(ctx, req.(*InspectContainerRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_ListServiceContainers_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(ListServiceContainersRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).ListServiceContainers(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_ListServiceContainers_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).ListServiceContainers(ctx, req.(*ListServiceContainersRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Docker_RemoveServiceContainer_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(RemoveContainerRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DockerServer).RemoveServiceContainer(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Docker_RemoveServiceContainer_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DockerServer).RemoveServiceContainer(ctx, req.(*RemoveContainerRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
 // Docker_ServiceDesc is the grpc.ServiceDesc for Docker service.
 // It's only intended for direct use with grpc.RegisterService,
 // and not to be introspected or modified (even as a copy)
@@ -342,6 +652,42 @@ var Docker_ServiceDesc = grpc.ServiceDesc{
 			MethodName: "RemoveContainer",
 			Handler:    _Docker_RemoveContainer_Handler,
 		},
+		{
+			MethodName: "InspectImage",
+			Handler:    _Docker_InspectImage_Handler,
+		},
+		{
+			MethodName: "InspectRemoteImage",
+			Handler:    _Docker_InspectRemoteImage_Handler,
+		},
+		{
+			MethodName: "CreateVolume",
+			Handler:    _Docker_CreateVolume_Handler,
+		},
+		{
+			MethodName: "ListVolumes",
+			Handler:    _Docker_ListVolumes_Handler,
+		},
+		{
+			MethodName: "RemoveVolume",
+			Handler:    _Docker_RemoveVolume_Handler,
+		},
+		{
+			MethodName: "CreateServiceContainer",
+			Handler:    _Docker_CreateServiceContainer_Handler,
+		},
+		{
+			MethodName: "InspectServiceContainer",
+			Handler:    _Docker_InspectServiceContainer_Handler,
+		},
+		{
+			MethodName: "ListServiceContainers",
+			Handler:    _Docker_ListServiceContainers_Handler,
+		},
+		{
+			MethodName: "RemoveServiceContainer",
+			Handler:    _Docker_RemoveServiceContainer_Handler,
+		},
 	},
 	Streams: []grpc.StreamDesc{
 		{

internal/machine/api/pb/machine.pb.go

@@ -163,6 +163,63 @@ func (x *NetworkConfig) GetPublicKey() []byte {
 	return nil
 }
 
+type CheckPrerequisitesResponse struct {
+	state         protoimpl.MessageState
+	sizeCache     protoimpl.SizeCache
+	unknownFields protoimpl.UnknownFields
+
+	// Overall status of the checks.
+	Satisfied bool `protobuf:"varint,1,opt,name=satisfied,proto3" json:"satisfied,omitempty"`
+	// Error message if not satisfied (empty if all checks pass).
+	Error string `protobuf:"bytes,2,opt,name=error,proto3" json:"error,omitempty"`
+}
+
+func (x *CheckPrerequisitesResponse) Reset() {
+	*x = CheckPrerequisitesResponse{}
+	if protoimpl.UnsafeEnabled {
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[2]
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		ms.StoreMessageInfo(mi)
+	}
+}
+
+func (x *CheckPrerequisitesResponse) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*CheckPrerequisitesResponse) ProtoMessage() {}
+
+func (x *CheckPrerequisitesResponse) ProtoReflect() protoreflect.Message {
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[2]
+	if protoimpl.UnsafeEnabled && x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use CheckPrerequisitesResponse.ProtoReflect.Descriptor instead.
+func (*CheckPrerequisitesResponse) Descriptor() ([]byte, []int) {
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{2}
+}
+
+func (x *CheckPrerequisitesResponse) GetSatisfied() bool {
+	if x != nil {
+		return x.Satisfied
+	}
+	return false
+}
+
+func (x *CheckPrerequisitesResponse) GetError() string {
+	if x != nil {
+		return x.Error
+	}
+	return ""
+}
+
 type InitClusterRequest struct {
 	state         protoimpl.MessageState
 	sizeCache     protoimpl.SizeCache
@@ -180,7 +237,7 @@ type InitClusterRequest struct {
 func (x *InitClusterRequest) Reset() {
 	*x = InitClusterRequest{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[2]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[3]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -193,7 +250,7 @@ func (x *InitClusterRequest) String() string {
 func (*InitClusterRequest) ProtoMessage() {}
 
 func (x *InitClusterRequest) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[2]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[3]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -206,7 +263,7 @@ func (x *InitClusterRequest) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use InitClusterRequest.ProtoReflect.Descriptor instead.
 func (*InitClusterRequest) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{2}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{3}
 }
 
 func (x *InitClusterRequest) GetMachineName() string {
@@ -271,7 +328,7 @@ type InitClusterResponse struct {
 func (x *InitClusterResponse) Reset() {
 	*x = InitClusterResponse{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[3]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[4]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -284,7 +341,7 @@ func (x *InitClusterResponse) String() string {
 func (*InitClusterResponse) ProtoMessage() {}
 
 func (x *InitClusterResponse) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[3]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[4]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -297,7 +354,7 @@ func (x *InitClusterResponse) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use InitClusterResponse.ProtoReflect.Descriptor instead.
 func (*InitClusterResponse) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{3}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{4}
 }
 
 func (x *InitClusterResponse) GetMachine() *MachineInfo {
@@ -319,7 +376,7 @@ type JoinClusterRequest struct {
 func (x *JoinClusterRequest) Reset() {
 	*x = JoinClusterRequest{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[4]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[5]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -332,7 +389,7 @@ func (x *JoinClusterRequest) String() string {
 func (*JoinClusterRequest) ProtoMessage() {}
 
 func (x *JoinClusterRequest) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[4]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[5]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -345,7 +402,7 @@ func (x *JoinClusterRequest) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use JoinClusterRequest.ProtoReflect.Descriptor instead.
 func (*JoinClusterRequest) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{4}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{5}
 }
 
 func (x *JoinClusterRequest) GetMachine() *MachineInfo {
@@ -373,7 +430,7 @@ type TokenResponse struct {
 func (x *TokenResponse) Reset() {
 	*x = TokenResponse{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[5]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[6]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -386,7 +443,7 @@ func (x *TokenResponse) String() string {
 func (*TokenResponse) ProtoMessage() {}
 
 func (x *TokenResponse) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[5]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[6]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -399,7 +456,7 @@ func (x *TokenResponse) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use TokenResponse.ProtoReflect.Descriptor instead.
 func (*TokenResponse) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{5}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{6}
 }
 
 func (x *TokenResponse) GetToken() string {
@@ -423,7 +480,7 @@ type Service struct {
 func (x *Service) Reset() {
 	*x = Service{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[6]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[7]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -436,7 +493,7 @@ func (x *Service) String() string {
 func (*Service) ProtoMessage() {}
 
 func (x *Service) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[6]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[7]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -449,7 +506,7 @@ func (x *Service) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use Service.ProtoReflect.Descriptor instead.
 func (*Service) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{6}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{7}
 }
 
 func (x *Service) GetId() string {
@@ -491,7 +548,7 @@ type InspectServiceRequest struct {
 func (x *InspectServiceRequest) Reset() {
 	*x = InspectServiceRequest{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[7]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[8]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -504,7 +561,7 @@ func (x *InspectServiceRequest) String() string {
 func (*InspectServiceRequest) ProtoMessage() {}
 
 func (x *InspectServiceRequest) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[7]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[8]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -517,7 +574,7 @@ func (x *InspectServiceRequest) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use InspectServiceRequest.ProtoReflect.Descriptor instead.
 func (*InspectServiceRequest) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{7}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{8}
 }
 
 func (x *InspectServiceRequest) GetId() string {
@@ -538,7 +595,7 @@ type InspectServiceResponse struct {
 func (x *InspectServiceResponse) Reset() {
 	*x = InspectServiceResponse{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[8]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[9]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -551,7 +608,7 @@ func (x *InspectServiceResponse) String() string {
 func (*InspectServiceResponse) ProtoMessage() {}
 
 func (x *InspectServiceResponse) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[8]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[9]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -564,7 +621,7 @@ func (x *InspectServiceResponse) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use InspectServiceResponse.ProtoReflect.Descriptor instead.
 func (*InspectServiceResponse) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{8}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{9}
 }
 
 func (x *InspectServiceResponse) GetService() *Service {
@@ -587,7 +644,7 @@ type Service_Container struct {
 func (x *Service_Container) Reset() {
 	*x = Service_Container{}
 	if protoimpl.UnsafeEnabled {
-		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[9]
+		mi := &file_internal_machine_api_pb_machine_proto_msgTypes[10]
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		ms.StoreMessageInfo(mi)
 	}
@@ -600,7 +657,7 @@ func (x *Service_Container) String() string {
 func (*Service_Container) ProtoMessage() {}
 
 func (x *Service_Container) ProtoReflect() protoreflect.Message {
-	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[9]
+	mi := &file_internal_machine_api_pb_machine_proto_msgTypes[10]
 	if protoimpl.UnsafeEnabled && x != nil {
 		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
 		if ms.LoadMessageInfo() == nil {
@@ -613,7 +670,7 @@ func (x *Service_Container) ProtoReflect() protoreflect.Message {
 
 // Deprecated: Use Service_Container.ProtoReflect.Descriptor instead.
 func (*Service_Container) Descriptor() ([]byte, []int) {
-	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{6, 0}
+	return file_internal_machine_api_pb_machine_proto_rawDescGZIP(), []int{7, 0}
 }
 
 func (x *Service_Container) GetMachineId() string {
@@ -659,77 +716,87 @@ var file_internal_machine_api_pb_machine_proto_rawDesc = []byte{
 	0x0b, 0x32, 0x0b, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x50, 0x50, 0x6f, 0x72, 0x74, 0x52, 0x09,
 	0x65, 0x6e, 0x64, 0x70, 0x6f, 0x69, 0x6e, 0x74, 0x73, 0x12, 0x1d, 0x0a, 0x0a, 0x70, 0x75, 0x62,
 	0x6c, 0x69, 0x63, 0x5f, 0x6b, 0x65, 0x79, 0x18, 0x04, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x09, 0x70,
-	0x75, 0x62, 0x6c, 0x69, 0x63, 0x4b, 0x65, 0x79, 0x22, 0xc3, 0x01, 0x0a, 0x12, 0x49, 0x6e, 0x69,
+	0x75, 0x62, 0x6c, 0x69, 0x63, 0x4b, 0x65, 0x79, 0x22, 0x50, 0x0a, 0x1a, 0x43, 0x68, 0x65, 0x63,
-	0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12,
+	0x6b, 0x50, 0x72, 0x65, 0x72, 0x65, 0x71, 0x75, 0x69, 0x73, 0x69, 0x74, 0x65, 0x73, 0x52, 0x65,
-	0x20, 0x0a, 0x0b, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x4e, 0x61, 0x6d, 0x65, 0x18, 0x01,
+	0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x1c, 0x0a, 0x09, 0x73, 0x61, 0x74, 0x69, 0x73, 0x66,
-	0x20, 0x01, 0x28, 0x09, 0x52, 0x0b, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x4e, 0x61, 0x6d,
+	0x69, 0x65, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x08, 0x52, 0x09, 0x73, 0x61, 0x74, 0x69, 0x73,
-	0x65, 0x12, 0x27, 0x0a, 0x07, 0x6e, 0x65, 0x74, 0x77, 0x6f, 0x72, 0x6b, 0x18, 0x02, 0x20, 0x01,
+	0x66, 0x69, 0x65, 0x64, 0x12, 0x14, 0x0a, 0x05, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x18, 0x02, 0x20,
-	0x28, 0x0b, 0x32, 0x0d, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x50, 0x50, 0x72, 0x65, 0x66, 0x69,
+	0x01, 0x28, 0x09, 0x52, 0x05, 0x65, 0x72, 0x72, 0x6f, 0x72, 0x22, 0xc3, 0x01, 0x0a, 0x12, 0x49,
-	0x78, 0x52, 0x07, 0x6e, 0x65, 0x74, 0x77, 0x6f, 0x72, 0x6b, 0x12, 0x26, 0x0a, 0x09, 0x70, 0x75,
+	0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73,
-	0x62, 0x6c, 0x69, 0x63, 0x5f, 0x69, 0x70, 0x18, 0x03, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x07, 0x2e,
+	0x74, 0x12, 0x20, 0x0a, 0x0b, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x4e, 0x61, 0x6d, 0x65,
-	0x61, 0x70, 0x69, 0x2e, 0x49, 0x50, 0x48, 0x00, 0x52, 0x08, 0x70, 0x75, 0x62, 0x6c, 0x69, 0x63,
+	0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x0b, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x4e,
-	0x49, 0x70, 0x12, 0x26, 0x0a, 0x0e, 0x70, 0x75, 0x62, 0x6c, 0x69, 0x63, 0x5f, 0x69, 0x70, 0x5f,
+	0x61, 0x6d, 0x65, 0x12, 0x27, 0x0a, 0x07, 0x6e, 0x65, 0x74, 0x77, 0x6f, 0x72, 0x6b, 0x18, 0x02,
-	0x61, 0x75, 0x74, 0x6f, 0x18, 0x04, 0x20, 0x01, 0x28, 0x08, 0x48, 0x00, 0x52, 0x0c, 0x70, 0x75,
+	0x20, 0x01, 0x28, 0x0b, 0x32, 0x0d, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x50, 0x50, 0x72, 0x65,
-	0x62, 0x6c, 0x69, 0x63, 0x49, 0x70, 0x41, 0x75, 0x74, 0x6f, 0x42, 0x12, 0x0a, 0x10, 0x70, 0x75,
+	0x66, 0x69, 0x78, 0x52, 0x07, 0x6e, 0x65, 0x74, 0x77, 0x6f, 0x72, 0x6b, 0x12, 0x26, 0x0a, 0x09,
-	0x62, 0x6c, 0x69, 0x63, 0x5f, 0x69, 0x70, 0x5f, 0x63, 0x6f, 0x6e, 0x66, 0x69, 0x67, 0x22, 0x41,
+	0x70, 0x75, 0x62, 0x6c, 0x69, 0x63, 0x5f, 0x69, 0x70, 0x18, 0x03, 0x20, 0x01, 0x28, 0x0b, 0x32,
-	0x0a, 0x13, 0x49, 0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72, 0x52, 0x65, 0x73,
+	0x07, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x50, 0x48, 0x00, 0x52, 0x08, 0x70, 0x75, 0x62, 0x6c,
-	0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x2a, 0x0a, 0x07, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65,
+	0x69, 0x63, 0x49, 0x70, 0x12, 0x26, 0x0a, 0x0e, 0x70, 0x75, 0x62, 0x6c, 0x69, 0x63, 0x5f, 0x69,
-	0x18, 0x01, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x10, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x4d, 0x61, 0x63,
+	0x70, 0x5f, 0x61, 0x75, 0x74, 0x6f, 0x18, 0x04, 0x20, 0x01, 0x28, 0x08, 0x48, 0x00, 0x52, 0x0c,
-	0x68, 0x69, 0x6e, 0x65, 0x49, 0x6e, 0x66, 0x6f, 0x52, 0x07, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e,
+	0x70, 0x75, 0x62, 0x6c, 0x69, 0x63, 0x49, 0x70, 0x41, 0x75, 0x74, 0x6f, 0x42, 0x12, 0x0a, 0x10,
-	0x65, 0x22, 0x79, 0x0a, 0x12, 0x4a, 0x6f, 0x69, 0x6e, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72,
+	0x70, 0x75, 0x62, 0x6c, 0x69, 0x63, 0x5f, 0x69, 0x70, 0x5f, 0x63, 0x6f, 0x6e, 0x66, 0x69, 0x67,
-	0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12, 0x2a, 0x0a, 0x07, 0x6d, 0x61, 0x63, 0x68, 0x69,
+	0x22, 0x41, 0x0a, 0x13, 0x49, 0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72, 0x52,
+	0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x2a, 0x0a, 0x07, 0x6d, 0x61, 0x63, 0x68, 0x69,
 	0x6e, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x10, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x4d,
 	0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x6e, 0x66, 0x6f, 0x52, 0x07, 0x6d, 0x61, 0x63, 0x68,
-	0x69, 0x6e, 0x65, 0x12, 0x37, 0x0a, 0x0e, 0x6f, 0x74, 0x68, 0x65, 0x72, 0x5f, 0x6d, 0x61, 0x63,
+	0x69, 0x6e, 0x65, 0x22, 0x79, 0x0a, 0x12, 0x4a, 0x6f, 0x69, 0x6e, 0x43, 0x6c, 0x75, 0x73, 0x74,
-	0x68, 0x69, 0x6e, 0x65, 0x73, 0x18, 0x03, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x10, 0x2e, 0x61, 0x70,
+	0x65, 0x72, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12, 0x2a, 0x0a, 0x07, 0x6d, 0x61, 0x63,
-	0x69, 0x2e, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x6e, 0x66, 0x6f, 0x52, 0x0d, 0x6f,
+	0x68, 0x69, 0x6e, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x10, 0x2e, 0x61, 0x70, 0x69,
-	0x74, 0x68, 0x65, 0x72, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x73, 0x22, 0x25, 0x0a, 0x0d,
+	0x2e, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x6e, 0x66, 0x6f, 0x52, 0x07, 0x6d, 0x61,
-	0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x14, 0x0a,
+	0x63, 0x68, 0x69, 0x6e, 0x65, 0x12, 0x37, 0x0a, 0x0e, 0x6f, 0x74, 0x68, 0x65, 0x72, 0x5f, 0x6d,
-	0x05, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x05, 0x74, 0x6f,
+	0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x73, 0x18, 0x03, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x10, 0x2e,
-	0x6b, 0x65, 0x6e, 0x22, 0xc3, 0x01, 0x0a, 0x07, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x12,
+	0x61, 0x70, 0x69, 0x2e, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x6e, 0x66, 0x6f, 0x52,
-	0x0e, 0x0a, 0x02, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x02, 0x69, 0x64, 0x12,
+	0x0d, 0x6f, 0x74, 0x68, 0x65, 0x72, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x73, 0x22, 0x25,
-	0x12, 0x0a, 0x04, 0x6e, 0x61, 0x6d, 0x65, 0x18, 0x02, 0x20, 0x01, 0x28, 0x09, 0x52, 0x04, 0x6e,
+	0x0a, 0x0d, 0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12,
-	0x61, 0x6d, 0x65, 0x12, 0x12, 0x0a, 0x04, 0x6d, 0x6f, 0x64, 0x65, 0x18, 0x03, 0x20, 0x01, 0x28,
+	0x14, 0x0a, 0x05, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x05,
-	0x09, 0x52, 0x04, 0x6d, 0x6f, 0x64, 0x65, 0x12, 0x36, 0x0a, 0x0a, 0x63, 0x6f, 0x6e, 0x74, 0x61,
+	0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x22, 0xc3, 0x01, 0x0a, 0x07, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63,
-	0x69, 0x6e, 0x65, 0x72, 0x73, 0x18, 0x04, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x16, 0x2e, 0x61, 0x70,
+	0x65, 0x12, 0x0e, 0x0a, 0x02, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x02, 0x69,
-	0x69, 0x2e, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x2e, 0x43, 0x6f, 0x6e, 0x74, 0x61, 0x69,
+	0x64, 0x12, 0x12, 0x0a, 0x04, 0x6e, 0x61, 0x6d, 0x65, 0x18, 0x02, 0x20, 0x01, 0x28, 0x09, 0x52,
-	0x6e, 0x65, 0x72, 0x52, 0x0a, 0x63, 0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x73, 0x1a,
+	0x04, 0x6e, 0x61, 0x6d, 0x65, 0x12, 0x12, 0x0a, 0x04, 0x6d, 0x6f, 0x64, 0x65, 0x18, 0x03, 0x20,
-	0x48, 0x0a, 0x09, 0x43, 0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x12, 0x1d, 0x0a, 0x0a,
+	0x01, 0x28, 0x09, 0x52, 0x04, 0x6d, 0x6f, 0x64, 0x65, 0x12, 0x36, 0x0a, 0x0a, 0x63, 0x6f, 0x6e,
-	0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x5f, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09,
+	0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x73, 0x18, 0x04, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x16, 0x2e,
-	0x52, 0x09, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x64, 0x12, 0x1c, 0x0a, 0x09, 0x63,
+	0x61, 0x70, 0x69, 0x2e, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x2e, 0x43, 0x6f, 0x6e, 0x74,
-	0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x09,
+	0x61, 0x69, 0x6e, 0x65, 0x72, 0x52, 0x0a, 0x63, 0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72,
-	0x63, 0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x22, 0x27, 0x0a, 0x15, 0x49, 0x6e, 0x73,
+	0x73, 0x1a, 0x48, 0x0a, 0x09, 0x43, 0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x12, 0x1d,
-	0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x52, 0x65, 0x71, 0x75, 0x65,
+	0x0a, 0x0a, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x5f, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01,
-	0x73, 0x74, 0x12, 0x0e, 0x0a, 0x02, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x02,
+	0x28, 0x09, 0x52, 0x09, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x64, 0x12, 0x1c, 0x0a,
-	0x69, 0x64, 0x22, 0x40, 0x0a, 0x16, 0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72,
+	0x09, 0x63, 0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0c,
-	0x76, 0x69, 0x63, 0x65, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x26, 0x0a, 0x07,
+	0x52, 0x09, 0x63, 0x6f, 0x6e, 0x74, 0x61, 0x69, 0x6e, 0x65, 0x72, 0x22, 0x27, 0x0a, 0x15, 0x49,
-	0x73, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x0c, 0x2e,
+	0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x52, 0x65, 0x71,
-	0x61, 0x70, 0x69, 0x2e, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x52, 0x07, 0x73, 0x65, 0x72,
+	0x75, 0x65, 0x73, 0x74, 0x12, 0x0e, 0x0a, 0x02, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09,
-	0x76, 0x69, 0x63, 0x65, 0x32, 0xc0, 0x02, 0x0a, 0x07, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65,
+	0x52, 0x02, 0x69, 0x64, 0x22, 0x40, 0x0a, 0x16, 0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53,
-	0x12, 0x40, 0x0a, 0x0b, 0x49, 0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72, 0x12,
+	0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x26,
-	0x17, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65,
+	0x0a, 0x07, 0x73, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0b, 0x32,
-	0x72, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x18, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49,
+	0x0c, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x52, 0x07, 0x73,
-	0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e,
+	0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x32, 0x8f, 0x03, 0x0a, 0x07, 0x4d, 0x61, 0x63, 0x68, 0x69,
-	0x73, 0x65, 0x12, 0x3e, 0x0a, 0x0b, 0x4a, 0x6f, 0x69, 0x6e, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65,
+	0x6e, 0x65, 0x12, 0x4d, 0x0a, 0x12, 0x43, 0x68, 0x65, 0x63, 0x6b, 0x50, 0x72, 0x65, 0x72, 0x65,
-	0x72, 0x12, 0x17, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x4a, 0x6f, 0x69, 0x6e, 0x43, 0x6c, 0x75, 0x73,
+	0x71, 0x75, 0x69, 0x73, 0x69, 0x74, 0x65, 0x73, 0x12, 0x16, 0x2e, 0x67, 0x6f, 0x6f, 0x67, 0x6c,
-	0x74, 0x65, 0x72, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x16, 0x2e, 0x67, 0x6f, 0x6f,
+	0x65, 0x2e, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x62, 0x75, 0x66, 0x2e, 0x45, 0x6d, 0x70, 0x74, 0x79,
-	0x67, 0x6c, 0x65, 0x2e, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x62, 0x75, 0x66, 0x2e, 0x45, 0x6d, 0x70,
+	0x1a, 0x1f, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x43, 0x68, 0x65, 0x63, 0x6b, 0x50, 0x72, 0x65, 0x72,
-	0x74, 0x79, 0x12, 0x33, 0x0a, 0x05, 0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x12, 0x16, 0x2e, 0x67, 0x6f,
+	0x65, 0x71, 0x75, 0x69, 0x73, 0x69, 0x74, 0x65, 0x73, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73,
+	0x65, 0x12, 0x40, 0x0a, 0x0b, 0x49, 0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72,
+	0x12, 0x17, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74,
+	0x65, 0x72, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x18, 0x2e, 0x61, 0x70, 0x69, 0x2e,
+	0x49, 0x6e, 0x69, 0x74, 0x43, 0x6c, 0x75, 0x73, 0x74, 0x65, 0x72, 0x52, 0x65, 0x73, 0x70, 0x6f,
+	0x6e, 0x73, 0x65, 0x12, 0x3e, 0x0a, 0x0b, 0x4a, 0x6f, 0x69, 0x6e, 0x43, 0x6c, 0x75, 0x73, 0x74,
+	0x65, 0x72, 0x12, 0x17, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x4a, 0x6f, 0x69, 0x6e, 0x43, 0x6c, 0x75,
+	0x73, 0x74, 0x65, 0x72, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x16, 0x2e, 0x67, 0x6f,
 	0x6f, 0x67, 0x6c, 0x65, 0x2e, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x62, 0x75, 0x66, 0x2e, 0x45, 0x6d,
-	0x70, 0x74, 0x79, 0x1a, 0x12, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x52,
+	0x70, 0x74, 0x79, 0x12, 0x33, 0x0a, 0x05, 0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x12, 0x16, 0x2e, 0x67,
-	0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x33, 0x0a, 0x07, 0x49, 0x6e, 0x73, 0x70, 0x65,
+	0x6f, 0x6f, 0x67, 0x6c, 0x65, 0x2e, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x62, 0x75, 0x66, 0x2e, 0x45,
-	0x63, 0x74, 0x12, 0x16, 0x2e, 0x67, 0x6f, 0x6f, 0x67, 0x6c, 0x65, 0x2e, 0x70, 0x72, 0x6f, 0x74,
+	0x6d, 0x70, 0x74, 0x79, 0x1a, 0x12, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x54, 0x6f, 0x6b, 0x65, 0x6e,
-	0x6f, 0x62, 0x75, 0x66, 0x2e, 0x45, 0x6d, 0x70, 0x74, 0x79, 0x1a, 0x10, 0x2e, 0x61, 0x70, 0x69,
+	0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x33, 0x0a, 0x07, 0x49, 0x6e, 0x73, 0x70,
-	0x2e, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x6e, 0x66, 0x6f, 0x12, 0x49, 0x0a, 0x0e,
+	0x65, 0x63, 0x74, 0x12, 0x16, 0x2e, 0x67, 0x6f, 0x6f, 0x67, 0x6c, 0x65, 0x2e, 0x70, 0x72, 0x6f,
-	0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x12, 0x1a,
+	0x74, 0x6f, 0x62, 0x75, 0x66, 0x2e, 0x45, 0x6d, 0x70, 0x74, 0x79, 0x1a, 0x10, 0x2e, 0x61, 0x70,
-	0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72, 0x76,
+	0x69, 0x2e, 0x4d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x49, 0x6e, 0x66, 0x6f, 0x12, 0x49, 0x0a,
-	0x69, 0x63, 0x65, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x1b, 0x2e, 0x61, 0x70, 0x69,
+	0x0e, 0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x12,
-	0x2e, 0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65, 0x52,
+	0x1a, 0x2e, 0x61, 0x70, 0x69, 0x2e, 0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72,
-	0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x42, 0x37, 0x5a, 0x35, 0x67, 0x69, 0x74, 0x68, 0x75,
+	0x76, 0x69, 0x63, 0x65, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x1b, 0x2e, 0x61, 0x70,
-	0x62, 0x2e, 0x63, 0x6f, 0x6d, 0x2f, 0x70, 0x73, 0x76, 0x69, 0x64, 0x65, 0x72, 0x73, 0x6b, 0x69,
+	0x69, 0x2e, 0x49, 0x6e, 0x73, 0x70, 0x65, 0x63, 0x74, 0x53, 0x65, 0x72, 0x76, 0x69, 0x63, 0x65,
-	0x2f, 0x75, 0x6e, 0x63, 0x6c, 0x6f, 0x75, 0x64, 0x2f, 0x69, 0x6e, 0x74, 0x65, 0x72, 0x6e, 0x61,
+	0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x42, 0x37, 0x5a, 0x35, 0x67, 0x69, 0x74, 0x68,
-	0x6c, 0x2f, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x2f, 0x61, 0x70, 0x69, 0x2f, 0x70, 0x62,
+	0x75, 0x62, 0x2e, 0x63, 0x6f, 0x6d, 0x2f, 0x70, 0x73, 0x76, 0x69, 0x64, 0x65, 0x72, 0x73, 0x6b,
-	0x62, 0x06, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x33,
+	0x69, 0x2f, 0x75, 0x6e, 0x63, 0x6c, 0x6f, 0x75, 0x64, 0x2f, 0x69, 0x6e, 0x74, 0x65, 0x72, 0x6e,
+	0x61, 0x6c, 0x2f, 0x6d, 0x61, 0x63, 0x68, 0x69, 0x6e, 0x65, 0x2f, 0x61, 0x70, 0x69, 0x2f, 0x70,
+	0x62, 0x62, 0x06, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x33,
 }
 
 var (
@@ -744,48 +811,51 @@ func file_internal_machine_api_pb_machine_proto_rawDescGZIP() []byte {
 	return file_internal_machine_api_pb_machine_proto_rawDescData
 }
 
-var file_internal_machine_api_pb_machine_proto_msgTypes = make([]protoimpl.MessageInfo, 10)
+var file_internal_machine_api_pb_machine_proto_msgTypes = make([]protoimpl.MessageInfo, 11)
 var file_internal_machine_api_pb_machine_proto_goTypes = []any{
-	(*MachineInfo)(nil),            // 0: api.MachineInfo
+	(*MachineInfo)(nil),                // 0: api.MachineInfo
-	(*NetworkConfig)(nil),          // 1: api.NetworkConfig
+	(*NetworkConfig)(nil),              // 1: api.NetworkConfig
-	(*InitClusterRequest)(nil),     // 2: api.InitClusterRequest
+	(*CheckPrerequisitesResponse)(nil), // 2: api.CheckPrerequisitesResponse
-	(*InitClusterResponse)(nil),    // 3: api.InitClusterResponse
+	(*InitClusterRequest)(nil),         // 3: api.InitClusterRequest
-	(*JoinClusterRequest)(nil),     // 4: api.JoinClusterRequest
+	(*InitClusterResponse)(nil),        // 4: api.InitClusterResponse
-	(*TokenResponse)(nil),          // 5: api.TokenResponse
+	(*JoinClusterRequest)(nil),         // 5: api.JoinClusterRequest
-	(*Service)(nil),                // 6: api.Service
+	(*TokenResponse)(nil),              // 6: api.TokenResponse
-	(*InspectServiceRequest)(nil),  // 7: api.InspectServiceRequest
+	(*Service)(nil),                    // 7: api.Service
-	(*InspectServiceResponse)(nil), // 8: api.InspectServiceResponse
+	(*InspectServiceRequest)(nil),      // 8: api.InspectServiceRequest
-	(*Service_Container)(nil),      // 9: api.Service.Container
+	(*InspectServiceResponse)(nil),     // 9: api.InspectServiceResponse
-	(*IP)(nil),                     // 10: api.IP
+	(*Service_Container)(nil),          // 10: api.Service.Container
-	(*IPPrefix)(nil),               // 11: api.IPPrefix
+	(*IP)(nil),                         // 11: api.IP
-	(*IPPort)(nil),                 // 12: api.IPPort
+	(*IPPrefix)(nil),                   // 12: api.IPPrefix
-	(*emptypb.Empty)(nil),          // 13: google.protobuf.Empty
+	(*IPPort)(nil),                     // 13: api.IPPort
+	(*emptypb.Empty)(nil),              // 14: google.protobuf.Empty
 }
 var file_internal_machine_api_pb_machine_proto_depIdxs = []int32{
 	1,  // 0: api.MachineInfo.network:type_name -> api.NetworkConfig
-	10, // 1: api.MachineInfo.public_ip:type_name -> api.IP
+	11, // 1: api.MachineInfo.public_ip:type_name -> api.IP
-	11, // 2: api.NetworkConfig.subnet:type_name -> api.IPPrefix
+	12, // 2: api.NetworkConfig.subnet:type_name -> api.IPPrefix
-	10, // 3: api.NetworkConfig.management_ip:type_name -> api.IP
+	11, // 3: api.NetworkConfig.management_ip:type_name -> api.IP
-	12, // 4: api.NetworkConfig.endpoints:type_name -> api.IPPort
+	13, // 4: api.NetworkConfig.endpoints:type_name -> api.IPPort
-	11, // 5: api.InitClusterRequest.network:type_name -> api.IPPrefix
+	12, // 5: api.InitClusterRequest.network:type_name -> api.IPPrefix
-	10, // 6: api.InitClusterRequest.public_ip:type_name -> api.IP
+	11, // 6: api.InitClusterRequest.public_ip:type_name -> api.IP
 	0,  // 7: api.InitClusterResponse.machine:type_name -> api.MachineInfo
 	0,  // 8: api.JoinClusterRequest.machine:type_name -> api.MachineInfo
 	0,  // 9: api.JoinClusterRequest.other_machines:type_name -> api.MachineInfo
-	9,  // 10: api.Service.containers:type_name -> api.Service.Container
+	10, // 10: api.Service.containers:type_name -> api.Service.Container
-	6,  // 11: api.InspectServiceResponse.service:type_name -> api.Service
+	7,  // 11: api.InspectServiceResponse.service:type_name -> api.Service
-	2,  // 12: api.Machine.InitCluster:input_type -> api.InitClusterRequest
+	14, // 12: api.Machine.CheckPrerequisites:input_type -> google.protobuf.Empty
-	4,  // 13: api.Machine.JoinCluster:input_type -> api.JoinClusterRequest
+	3,  // 13: api.Machine.InitCluster:input_type -> api.InitClusterRequest
-	13, // 14: api.Machine.Token:input_type -> google.protobuf.Empty
+	5,  // 14: api.Machine.JoinCluster:input_type -> api.JoinClusterRequest
-	13, // 15: api.Machine.Inspect:input_type -> google.protobuf.Empty
+	14, // 15: api.Machine.Token:input_type -> google.protobuf.Empty
-	7,  // 16: api.Machine.InspectService:input_type -> api.InspectServiceRequest
+	14, // 16: api.Machine.Inspect:input_type -> google.protobuf.Empty
-	3,  // 17: api.Machine.InitCluster:output_type -> api.InitClusterResponse
+	8,  // 17: api.Machine.InspectService:input_type -> api.InspectServiceRequest
-	13, // 18: api.Machine.JoinCluster:output_type -> google.protobuf.Empty
+	2,  // 18: api.Machine.CheckPrerequisites:output_type -> api.CheckPrerequisitesResponse
-	5,  // 19: api.Machine.Token:output_type -> api.TokenResponse
+	4,  // 19: api.Machine.InitCluster:output_type -> api.InitClusterResponse
-	0,  // 20: api.Machine.Inspect:output_type -> api.MachineInfo
+	14, // 20: api.Machine.JoinCluster:output_type -> google.protobuf.Empty
-	8,  // 21: api.Machine.InspectService:output_type -> api.InspectServiceResponse
+	6,  // 21: api.Machine.Token:output_type -> api.TokenResponse
-	17, // [17:22] is the sub-list for method output_type
+	0,  // 22: api.Machine.Inspect:output_type -> api.MachineInfo
-	12, // [12:17] is the sub-list for method input_type
+	9,  // 23: api.Machine.InspectService:output_type -> api.InspectServiceResponse
+	18, // [18:24] is the sub-list for method output_type
+	12, // [12:18] is the sub-list for method input_type
 	12, // [12:12] is the sub-list for extension type_name
 	12, // [12:12] is the sub-list for extension extendee
 	0,  // [0:12] is the sub-list for field type_name
@@ -823,7 +893,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[2].Exporter = func(v any, i int) any {
-			switch v := v.(*InitClusterRequest); i {
+			switch v := v.(*CheckPrerequisitesResponse); i {
 			case 0:
 				return &v.state
 			case 1:
@@ -835,7 +905,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[3].Exporter = func(v any, i int) any {
-			switch v := v.(*InitClusterResponse); i {
+			switch v := v.(*InitClusterRequest); i {
 			case 0:
 				return &v.state
 			case 1:
@@ -847,7 +917,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[4].Exporter = func(v any, i int) any {
-			switch v := v.(*JoinClusterRequest); i {
+			switch v := v.(*InitClusterResponse); i {
 			case 0:
 				return &v.state
 			case 1:
@@ -859,7 +929,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[5].Exporter = func(v any, i int) any {
-			switch v := v.(*TokenResponse); i {
+			switch v := v.(*JoinClusterRequest); i {
 			case 0:
 				return &v.state
 			case 1:
@@ -871,7 +941,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[6].Exporter = func(v any, i int) any {
-			switch v := v.(*Service); i {
+			switch v := v.(*TokenResponse); i {
 			case 0:
 				return &v.state
 			case 1:
@@ -883,7 +953,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[7].Exporter = func(v any, i int) any {
-			switch v := v.(*InspectServiceRequest); i {
+			switch v := v.(*Service); i {
 			case 0:
 				return &v.state
 			case 1:
@@ -895,7 +965,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[8].Exporter = func(v any, i int) any {
-			switch v := v.(*InspectServiceResponse); i {
+			switch v := v.(*InspectServiceRequest); i {
 			case 0:
 				return &v.state
 			case 1:
@@ -907,6 +977,18 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 		file_internal_machine_api_pb_machine_proto_msgTypes[9].Exporter = func(v any, i int) any {
+			switch v := v.(*InspectServiceResponse); i {
+			case 0:
+				return &v.state
+			case 1:
+				return &v.sizeCache
+			case 2:
+				return &v.unknownFields
+			default:
+				return nil
+			}
+		}
+		file_internal_machine_api_pb_machine_proto_msgTypes[10].Exporter = func(v any, i int) any {
 			switch v := v.(*Service_Container); i {
 			case 0:
 				return &v.state
@@ -919,7 +1001,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			}
 		}
 	}
-	file_internal_machine_api_pb_machine_proto_msgTypes[2].OneofWrappers = []any{
+	file_internal_machine_api_pb_machine_proto_msgTypes[3].OneofWrappers = []any{
 		(*InitClusterRequest_PublicIp)(nil),
 		(*InitClusterRequest_PublicIpAuto)(nil),
 	}
@@ -929,7 +1011,7 @@ func file_internal_machine_api_pb_machine_proto_init() {
 			GoPackagePath: reflect.TypeOf(x{}).PkgPath(),
 			RawDescriptor: file_internal_machine_api_pb_machine_proto_rawDesc,
 			NumEnums:      0,
-			NumMessages:   10,
+			NumMessages:   11,
 			NumExtensions: 0,
 			NumServices:   1,
 		},

internal/machine/api/pb/machine.proto

@@ -8,6 +8,8 @@ import "google/protobuf/empty.proto";
 import "internal/machine/api/pb/common.proto";
 
 service Machine {
+  // CheckPrerequisites verifies if the machine meets all necessary system requirements to participate in the cluster.
+  rpc CheckPrerequisites(google.protobuf.Empty) returns (CheckPrerequisitesResponse);
   rpc InitCluster(InitClusterRequest) returns (InitClusterResponse);
   rpc JoinCluster(JoinClusterRequest) returns (google.protobuf.Empty);
   rpc Token(google.protobuf.Empty) returns (TokenResponse);
@@ -29,6 +31,13 @@ message NetworkConfig {
   bytes public_key = 4;
 }
 
+message CheckPrerequisitesResponse {
+  // Overall status of the checks.
+  bool satisfied = 1;
+  // Error message if not satisfied (empty if all checks pass).
+  string error = 2;
+}
+
 message InitClusterRequest {
   string machineName = 1;
   IPPrefix network = 2;

internal/machine/api/pb/machine_grpc.pb.go

@@ -20,17 +20,20 @@ import (
 const _ = grpc.SupportPackageIsVersion9
 
 const (
-	Machine_InitCluster_FullMethodName    = "/api.Machine/InitCluster"
+	Machine_CheckPrerequisites_FullMethodName = "/api.Machine/CheckPrerequisites"
-	Machine_JoinCluster_FullMethodName    = "/api.Machine/JoinCluster"
+	Machine_InitCluster_FullMethodName        = "/api.Machine/InitCluster"
-	Machine_Token_FullMethodName          = "/api.Machine/Token"
+	Machine_JoinCluster_FullMethodName        = "/api.Machine/JoinCluster"
-	Machine_Inspect_FullMethodName        = "/api.Machine/Inspect"
+	Machine_Token_FullMethodName              = "/api.Machine/Token"
-	Machine_InspectService_FullMethodName = "/api.Machine/InspectService"
+	Machine_Inspect_FullMethodName            = "/api.Machine/Inspect"
+	Machine_InspectService_FullMethodName     = "/api.Machine/InspectService"
 )
 
 // MachineClient is the client API for Machine service.
 //
 // For semantics around ctx use and closing/ending streaming RPCs, please refer to https://pkg.go.dev/google.golang.org/grpc/?tab=doc#ClientConn.NewStream.
 type MachineClient interface {
+	// CheckPrerequisites verifies if the machine meets all necessary system requirements to participate in the cluster.
+	CheckPrerequisites(ctx context.Context, in *emptypb.Empty, opts ...grpc.CallOption) (*CheckPrerequisitesResponse, error)
 	InitCluster(ctx context.Context, in *InitClusterRequest, opts ...grpc.CallOption) (*InitClusterResponse, error)
 	JoinCluster(ctx context.Context, in *JoinClusterRequest, opts ...grpc.CallOption) (*emptypb.Empty, error)
 	Token(ctx context.Context, in *emptypb.Empty, opts ...grpc.CallOption) (*TokenResponse, error)
@@ -46,6 +49,16 @@ func NewMachineClient(cc grpc.ClientConnInterface) MachineClient {
 	return &machineClient{cc}
 }
 
+func (c *machineClient) CheckPrerequisites(ctx context.Context, in *emptypb.Empty, opts ...grpc.CallOption) (*CheckPrerequisitesResponse, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(CheckPrerequisitesResponse)
+	err := c.cc.Invoke(ctx, Machine_CheckPrerequisites_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
 func (c *machineClient) InitCluster(ctx context.Context, in *InitClusterRequest, opts ...grpc.CallOption) (*InitClusterResponse, error) {
 	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
 	out := new(InitClusterResponse)
@@ -100,6 +113,8 @@ func (c *machineClient) InspectService(ctx context.Context, in *InspectServiceRe
 // All implementations must embed UnimplementedMachineServer
 // for forward compatibility.
 type MachineServer interface {
+	// CheckPrerequisites verifies if the machine meets all necessary system requirements to participate in the cluster.
+	CheckPrerequisites(context.Context, *emptypb.Empty) (*CheckPrerequisitesResponse, error)
 	InitCluster(context.Context, *InitClusterRequest) (*InitClusterResponse, error)
 	JoinCluster(context.Context, *JoinClusterRequest) (*emptypb.Empty, error)
 	Token(context.Context, *emptypb.Empty) (*TokenResponse, error)
@@ -115,6 +130,9 @@ type MachineServer interface {
 // pointer dereference when methods are called.
 type UnimplementedMachineServer struct{}
 
+func (UnimplementedMachineServer) CheckPrerequisites(context.Context, *emptypb.Empty) (*CheckPrerequisitesResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method CheckPrerequisites not implemented")
+}
 func (UnimplementedMachineServer) InitCluster(context.Context, *InitClusterRequest) (*InitClusterResponse, error) {
 	return nil, status.Errorf(codes.Unimplemented, "method InitCluster not implemented")
 }
@@ -151,6 +169,24 @@ func RegisterMachineServer(s grpc.ServiceRegistrar, srv MachineServer) {
 	s.RegisterService(&Machine_ServiceDesc, srv)
 }
 
+func _Machine_CheckPrerequisites_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(emptypb.Empty)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(MachineServer).CheckPrerequisites(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: Machine_CheckPrerequisites_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(MachineServer).CheckPrerequisites(ctx, req.(*emptypb.Empty))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
 func _Machine_InitCluster_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
 	in := new(InitClusterRequest)
 	if err := dec(in); err != nil {
@@ -248,6 +284,10 @@ var Machine_ServiceDesc = grpc.ServiceDesc{
 	ServiceName: "api.Machine",
 	HandlerType: (*MachineServer)(nil),
 	Methods: []grpc.MethodDesc{
+		{
+			MethodName: "CheckPrerequisites",
+			Handler:    _Machine_CheckPrerequisites_Handler,
+		},
 		{
 			MethodName: "InitCluster",
 			Handler:    _Machine_InitCluster_Handler,

internal/machine/api/proxy/backend.go

@@ -5,7 +5,7 @@ import (
 	"google.golang.org/grpc/status"
 	"google.golang.org/protobuf/encoding/protowire"
 	"google.golang.org/protobuf/proto"
-	"uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
 )
 
 // One2ManyResponder converts upstream responses into messages from upstreams, so that multiple

internal/machine/caddyconfig/config.go

@@ -1,56 +1,61 @@
-package caddyfile
+package caddyconfig
 
 import (
 	"encoding/json"
 	"errors"
 	"fmt"
-	"github.com/caddyserver/caddy/v2"
-	"github.com/caddyserver/caddy/v2/caddyconfig"
-	"github.com/caddyserver/caddy/v2/modules/caddyhttp"
-	"github.com/caddyserver/caddy/v2/modules/caddyhttp/reverseproxy"
 	"log/slog"
 	"maps"
 	"net"
 	"net/http"
 	"slices"
 	"strconv"
-	"uncloud/internal/api"
+	"time"
-	"uncloud/internal/machine/docker"
+
+	"github.com/caddyserver/caddy/v2"
+	"github.com/caddyserver/caddy/v2/caddyconfig"
+	"github.com/caddyserver/caddy/v2/modules/caddyhttp"
+	"github.com/caddyserver/caddy/v2/modules/caddyhttp/reverseproxy"
+	"github.com/psviderski/uncloud/pkg/api"
 )
 
-func GenerateConfig(containers []api.Container, verifyResponse string) (*caddy.Config, error) {
+func GenerateConfig(containers []api.ServiceContainer, verifyResponse string) (*caddy.Config, error) {
 	// Maps hostnames to lists of upstreams (container IP:port pairs).
 	httpHostUpstreams := make(map[string][]string)
 	httpsHostUpstreams := make(map[string][]string)
 	for _, ctr := range containers {
-		logger := slog.With("container", ctr.ID)
+		if !ctr.Healthy() {
-		network, ok := ctr.NetworkSettings.Networks[docker.NetworkName]
-		if !ok {
-			// Container is not connected to the uncloud Docker network (could be host network).
-			continue
-		}
-		if network.IPAddress == "" {
-			logger.Error("Container has no IPv4 address.")
 			continue
 		}
 
+		ip := ctr.UncloudNetworkIP()
+		if !ip.IsValid() {
+			// Container is not connected to the uncloud Docker network (could be host network).
+			continue
+		}
+		log := slog.With("container", ctr.ID)
+
 		ports, err := ctr.ServicePorts()
 		if err != nil {
-			logger.Error("Failed to parse service ports for container.", "err", err)
+			log.Error("Failed to parse service ports for container.", "err", err)
 			continue
 		}
 
 		for _, port := range ports {
+			if port.Mode != api.PortModeIngress {
+				continue
+			}
+
 			switch port.Protocol {
 			case api.ProtocolHTTP:
-				upstream := net.JoinHostPort(network.IPAddress, strconv.Itoa(int(port.ContainerPort)))
+				upstream := net.JoinHostPort(ip.String(), strconv.Itoa(int(port.ContainerPort)))
 				httpHostUpstreams[port.Hostname] = append(httpHostUpstreams[port.Hostname], upstream)
 			case api.ProtocolHTTPS:
-				upstream := net.JoinHostPort(network.IPAddress, strconv.Itoa(int(port.ContainerPort)))
+				upstream := net.JoinHostPort(ip.String(), strconv.Itoa(int(port.ContainerPort)))
 				httpsHostUpstreams[port.Hostname] = append(httpsHostUpstreams[port.Hostname], upstream)
 			default:
 				// TODO: implement L4 ingress routing for TCP and UDP.
-				logger.Error("Unsupported protocol for ingress port.", "port", port)
+				log.Error("Unsupported protocol for ingress port.", "port", port)
 				continue
 			}
 		}
@@ -60,6 +65,8 @@ func GenerateConfig(containers []api.Container, verifyResponse string) (*caddy.C
 	servers := make(map[string]*caddyhttp.Server)
 	servers["http"] = &caddyhttp.Server{
 		Listen: []string{fmt.Sprintf(":%d", caddyhttp.DefaultHTTPPort)},
+		// All http requests to this server are logged to the default logger.
+		Logs: &caddyhttp.ServerLogConfig{},
 		Routes: append(
 			hostUpstreamsToRoutes(httpHostUpstreams, &warnings),
 			// Add a route to respond with a static verification response at the /.uncloud-verify path.
@@ -68,6 +75,8 @@ func GenerateConfig(containers []api.Container, verifyResponse string) (*caddy.C
 	}
 	servers["https"] = &caddyhttp.Server{
 		Listen: []string{fmt.Sprintf(":%d", caddyhttp.DefaultHTTPSPort)},
+		// All https requests to this server are logged to the default logger.
+		Logs:   &caddyhttp.ServerLogConfig{},
 		Routes: hostUpstreamsToRoutes(httpsHostUpstreams, &warnings),
 	}
 
@@ -108,6 +117,16 @@ func hostUpstreamsToRoutes(hostUpstreams map[string][]string, warnings *[]caddyc
 			}
 		}
 		handler := &reverseproxy.Handler{
+			HealthChecks: &reverseproxy.HealthChecks{
+				// Enable passive health checks to automatically detect unhealthy upstreams.
+				Passive: &reverseproxy.PassiveHealthChecks{
+					FailDuration: caddy.Duration(30 * time.Second),
+				},
+			},
+			LoadBalancing: &reverseproxy.LoadBalancing{
+				// Retry failed requests to skip over temporarily unavailable upstreams.
+				Retries: 3,
+			},
 			Upstreams: upstreamPool,
 		}
 

internal/machine/caddyconfig/config_test.go

@@ -1,51 +1,56 @@
-package caddyfile
+package caddyconfig
 
 import (
+	"strings"
+	"testing"
+
 	"github.com/docker/docker/api/types"
 	"github.com/docker/docker/api/types/container"
 	"github.com/docker/docker/api/types/network"
+	"github.com/psviderski/uncloud/internal/machine/docker"
+	"github.com/psviderski/uncloud/pkg/api"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
-	"strings"
-	"testing"
-	"uncloud/internal/api"
-	"uncloud/internal/machine/docker"
 )
 
 func TestGenerateConfig(t *testing.T) {
+	configWithoutServices := `{
+		"servers": {
+			"http": {
+				"listen": [":80"],
+				"routes": [{
+					"match": [{"path": ["/.uncloud-verify"]}],
+					"handle": [{
+						"body": "verification-response-body",
+						"handler": "static_response",
+						"status_code": 200
+					}]
+				}],
+				"logs": {}
+			},
+			"https": {
+				"listen": [":443"],
+				"logs": {}
+			}
+		}
+	}`
+
 	tests := []struct {
 		name       string
-		containers []api.Container
+		containers []api.ServiceContainer
 		want       string
 		wantErr    bool
 	}{
 		{
 			name:       "empty containers",
-			containers: []api.Container{},
+			containers: []api.ServiceContainer{},
-			want: `{
+			want:       configWithoutServices,
-                "servers": {
+			wantErr:    false,
-                    "http": {
-                        "listen": [":80"],
-                        "routes": [{
-                            "match": [{"path": ["/.uncloud-verify"]}],
-                            "handle": [{
-								"body": "verification-response-body",
-								"handler": "static_response",
-								"status_code": 200
-							}]
-                        }]
-                    },
-                    "https": {
-                        "listen": [":443"]
-                    }
-				}
-			}`,
-			wantErr: false,
 		},
 
 		{
 			name: "HTTP container",
-			containers: []api.Container{
+			containers: []api.ServiceContainer{
 				newContainer("10.210.0.2", "app.example.com:8080/http"),
 			},
 			want: `{
@@ -57,6 +62,14 @@ func TestGenerateConfig(t *testing.T) {
                                 "match": [{"host": ["app.example.com"]}],
                                 "handle": [{
                                     "handler": "reverse_proxy",
+									"health_checks": {
+                                        "passive": {
+                                            "fail_duration": 30000000000
+                                        }
+                                    },
+                                    "load_balancing": {
+                                        "retries": 3
+                                    },
                                     "upstreams": [{"dial": "10.210.0.2:8080"}]
                                 }]
                             },
@@ -68,10 +81,12 @@ func TestGenerateConfig(t *testing.T) {
                                     "status_code": 200
                                 }]
                             }
-                        ]
+                        ],
+						"logs": {}
                     },
                     "https": {
-                        "listen": [":443"]
+                        "listen": [":443"],
+						"logs": {}
                     }
                 }
             }`,
@@ -79,7 +94,7 @@ func TestGenerateConfig(t *testing.T) {
 		},
 		{
 			name: "load balancing multiple containers",
-			containers: []api.Container{
+			containers: []api.ServiceContainer{
 				newContainer("10.210.0.2", "app.example.com:8080/http"),
 				newContainer("10.210.0.3", "app.example.com:8080/http"),
 			},
@@ -92,6 +107,14 @@ func TestGenerateConfig(t *testing.T) {
                                 "match": [{"host": ["app.example.com"]}],
                                 "handle": [{
                                     "handler": "reverse_proxy",
+									"health_checks": {
+                                        "passive": {
+                                            "fail_duration": 30000000000
+                                        }
+                                    },
+                                    "load_balancing": {
+                                        "retries": 3
+                                    },
                                     "upstreams": [
                                         {"dial": "10.210.0.2:8080"},
                                         {"dial": "10.210.0.3:8080"}
@@ -106,10 +129,12 @@ func TestGenerateConfig(t *testing.T) {
                                     "status_code": 200
                                 }]
                             }
-                        ]
+                        ],
+						"logs": {}
                     },
                     "https": {
-                        "listen": [":443"]
+                        "listen": [":443"],
+						"logs": {}
                     }
                 }
             }`,
@@ -117,7 +142,7 @@ func TestGenerateConfig(t *testing.T) {
 		},
 		{
 			name: "HTTPS container",
-			containers: []api.Container{
+			containers: []api.ServiceContainer{
 				newContainer("10.210.0.2", "secure.example.com:8000/https"),
 			},
 			want: `{
@@ -133,7 +158,8 @@ func TestGenerateConfig(t *testing.T) {
                                     "status_code": 200
                                 }]
                             }
-                        ]
+                        ],
+						"logs": {}
                     },
                     "https": {
                         "listen": [":443"],
@@ -142,10 +168,19 @@ func TestGenerateConfig(t *testing.T) {
                                 "match": [{"host": ["secure.example.com"]}],
                                 "handle": [{
                                     "handler": "reverse_proxy",
+									"health_checks": {
+                                        "passive": {
+                                            "fail_duration": 30000000000
+                                        }
+                                    },
+                                    "load_balancing": {
+                                        "retries": 3
+                                    },
                                     "upstreams": [{"dial": "10.210.0.2:8000"}]
                                 }]
                             }
-                        ]
+                        ],
+						"logs": {}
                     }
                 }
             }`,
@@ -153,7 +188,7 @@ func TestGenerateConfig(t *testing.T) {
 		},
 		{
 			name: "mixed HTTP and HTTPS",
-			containers: []api.Container{
+			containers: []api.ServiceContainer{
 				newContainer("10.210.0.2",
 					"app.example.com:8080/http",
 					"web.example.com:8000/http"),
@@ -177,6 +212,14 @@ func TestGenerateConfig(t *testing.T) {
                                 "match": [{"host": ["app.example.com"]}],
                                 "handle": [{
                                     "handler": "reverse_proxy",
+									"health_checks": {
+                                        "passive": {
+                                            "fail_duration": 30000000000
+                                        }
+                                    },
+                                    "load_balancing": {
+                                        "retries": 3
+                                    },
                                     "upstreams": [
 										{"dial": "10.210.0.2:8080"},
 										{"dial": "10.210.0.3:8080"},
@@ -188,6 +231,14 @@ func TestGenerateConfig(t *testing.T) {
                                 "match": [{"host": ["web.example.com"]}],
                                 "handle": [{
                                     "handler": "reverse_proxy",
+									"health_checks": {
+                                        "passive": {
+                                            "fail_duration": 30000000000
+                                        }
+                                    },
+                                    "load_balancing": {
+                                        "retries": 3
+                                    },
                                     "upstreams": [
 										{"dial": "10.210.0.2:8000"},
 										{"dial": "10.210.0.4:8000"},
@@ -203,7 +254,8 @@ func TestGenerateConfig(t *testing.T) {
                                     "status_code": 200
                                 }]
                             }
-                        ]
+                        ],
+						"logs": {}
                     },
                     "https": {
                         "listen": [":443"],
@@ -212,6 +264,14 @@ func TestGenerateConfig(t *testing.T) {
                                 "match": [{"host": ["secure.example.com"]}],
                                 "handle": [{
                                     "handler": "reverse_proxy",
+									"health_checks": {
+                                        "passive": {
+                                            "fail_duration": 30000000000
+                                        }
+                                    },
+                                    "load_balancing": {
+                                        "retries": 3
+                                    },
                                     "upstreams": [
 										{"dial": "10.210.0.3:8888"},
 										{"dial": "10.210.0.4:8888"},
@@ -219,7 +279,8 @@ func TestGenerateConfig(t *testing.T) {
 									]
                                 }]
                             }
-                        ]
+                        ],
+						"logs": {}
                     }
                 }
             }`,
@@ -227,76 +288,87 @@ func TestGenerateConfig(t *testing.T) {
 		},
 		{
 			name: "container without uncloud network ignored",
-			containers: []api.Container{
+			containers: []api.ServiceContainer{
 				newContainerWithoutNetwork("ignored.example.com:8080/http"),
 			},
-			want: `{
+			want:    configWithoutServices,
-                "servers": {
-                    "http": {
-                        "listen": [":80"],
-                        "routes": [{
-                            "match": [{"path": ["/.uncloud-verify"]}],
-                            "handle": [{
-                                "body": "verification-response-body",
-                                "handler": "static_response",
-                                "status_code": 200
-                            }]
-                        }]
-                    },
-                    "https": {
-                        "listen": [":443"]
-                    }
-                }
-            }`,
 			wantErr: false,
 		},
 		{
 			name: "container with invalid port ignored",
-			containers: []api.Container{
+			containers: []api.ServiceContainer{
 				newContainer("10.210.0.2", "invalid-port"),
 			},
-			want: `{
+			want:    configWithoutServices,
-                "servers": {
-                    "http": {
-                        "listen": [":80"],
-                        "routes": [{
-                            "match": [{"path": ["/.uncloud-verify"]}],
-                            "handle": [{
-                                "body": "verification-response-body",
-                                "handler": "static_response",
-                                "status_code": 200
-                            }]
-                        }]
-                    },
-                    "https": {
-                        "listen": [":443"]
-                    }
-                }
-            }`,
 			wantErr: false,
 		},
 		{
-			name: "containers with unsupported protocols ignored",
+			name: "containers with unsupported protocols and host mode ignored",
-			containers: []api.Container{
+			containers: []api.ServiceContainer{
 				newContainer("10.210.0.2", "5000/tcp"),
 				newContainer("10.210.0.3", "5000/udp"),
 				newContainer("10.210.0.4", "80:8080/tcp@host"),
 			},
+			want:    configWithoutServices,
+			wantErr: false,
+		},
+		{
+			name: "restarting container ignored",
+			containers: []api.ServiceContainer{
+				newRestartingContainer("10.210.0.2", "app.example.com:8080/http"),
+			},
+			want:    configWithoutServices,
+			wantErr: false,
+		},
+		{
+			name: "stopped container ignored",
+			containers: []api.ServiceContainer{
+				newStoppedContainer("10.210.0.2", "app.example.com:8080/http"),
+			},
+			want:    configWithoutServices,
+			wantErr: false,
+		},
+		{
+			name: "mix of running, restarting, and stopped containers",
+			containers: []api.ServiceContainer{
+				newContainer("10.210.0.2", "app.example.com:8080/http"),
+				newRestartingContainer("10.210.0.3", "app.example.com:8080/http"),
+				newStoppedContainer("10.210.0.4", "app.example.com:8080/http"),
+			},
 			want: `{
                 "servers": {
                     "http": {
                         "listen": [":80"],
-                        "routes": [{
+                        "routes": [
-                            "match": [{"path": ["/.uncloud-verify"]}],
+                            {
-                            "handle": [{
+                                "match": [{"host": ["app.example.com"]}],
-                                "body": "verification-response-body",
+                                "handle": [{
-                                "handler": "static_response",
+                                    "handler": "reverse_proxy",
-                                "status_code": 200
+									"health_checks": {
-                            }]
+                                        "passive": {
-                        }]
+                                            "fail_duration": 30000000000
+                                        }
+                                    },
+                                    "load_balancing": {
+                                        "retries": 3
+                                    },
+                                    "upstreams": [{"dial": "10.210.0.2:8080"}]
+                                }]
+                            },
+                            {
+                                "match": [{"path": ["/.uncloud-verify"]}],
+                                "handle": [{
+                                    "body": "verification-response-body",
+                                    "handler": "static_response",
+                                    "status_code": 200
+                                }]
+                            }
+                        ],
+						"logs": {}
                     },
                     "https": {
-                        "listen": [":443"]
+                        "listen": [":443"],
+						"logs": {}
                     }
                 }
             }`,
@@ -322,44 +394,60 @@ func TestGenerateConfig(t *testing.T) {
 	}
 }
 
-func newContainer(ip string, ports ...string) api.Container {
+func newContainer(ip string, ports ...string) api.ServiceContainer {
 	portsLabel := strings.Join(ports, ",")
-	return api.Container{
+	return api.ServiceContainer{Container: api.Container{ContainerJSON: types.ContainerJSON{
-		ContainerJSON: types.ContainerJSON{
+		ContainerJSONBase: &types.ContainerJSONBase{
-			ContainerJSONBase: &types.ContainerJSONBase{},
+			State: &types.ContainerState{
-			NetworkSettings: &types.NetworkSettings{
+				Running: true,
-				Networks: map[string]*network.EndpointSettings{
-					docker.NetworkName: {
-						IPAddress: ip,
-					},
-				},
 			},
-			Config: &container.Config{
+		},
-				Labels: map[string]string{
+		NetworkSettings: &types.NetworkSettings{
-					api.LabelServicePorts: portsLabel,
+			Networks: map[string]*network.EndpointSettings{
+				docker.NetworkName: {
+					IPAddress: ip,
 				},
 			},
 		},
-	}
+		Config: &container.Config{
+			Labels: map[string]string{
+				api.LabelServicePorts: portsLabel,
+			},
+		},
+	}}}
 }
 
-func newContainerWithoutNetwork(ports ...string) api.Container {
+func newContainerWithoutNetwork(ports ...string) api.ServiceContainer {
 	portsLabel := strings.Join(ports, ",")
-	return api.Container{
+	return api.ServiceContainer{Container: api.Container{ContainerJSON: types.ContainerJSON{
-		ContainerJSON: types.ContainerJSON{
+		ContainerJSONBase: &types.ContainerJSONBase{
-			ContainerJSONBase: &types.ContainerJSONBase{},
+			State: &types.ContainerState{
-			NetworkSettings: &types.NetworkSettings{
+				Running: true,
-				Networks: map[string]*network.EndpointSettings{
-					"other-network": {
-						IPAddress: "172.17.0.2",
-					},
-				},
 			},
-			Config: &container.Config{
+		},
-				Labels: map[string]string{
+		NetworkSettings: &types.NetworkSettings{
-					api.LabelServicePorts: portsLabel,
+			Networks: map[string]*network.EndpointSettings{
+				"other-network": {
+					IPAddress: "172.17.0.2",
 				},
 			},
 		},
-	}
+		Config: &container.Config{
+			Labels: map[string]string{
+				api.LabelServicePorts: portsLabel,
+			},
+		},
+	}}}
+}
+
+func newRestartingContainer(ip string, ports ...string) api.ServiceContainer {
+	ctr := newContainer(ip, ports...)
+	ctr.Container.State.Restarting = true
+	return ctr
+}
+
+func newStoppedContainer(ip string, ports ...string) api.ServiceContainer {
+	ctr := newContainer(ip, ports...)
+	ctr.Container.State.Running = false
+	return ctr
 }

internal/machine/caddyconfig/controller.go

@@ -1,4 +1,4 @@
-package caddyfile
+package caddyconfig
 
 import (
 	"context"
@@ -7,9 +7,10 @@ import (
 	"log/slog"
 	"os"
 	"path/filepath"
-	"uncloud/internal/api"
+
-	"uncloud/internal/fs"
+	"github.com/psviderski/uncloud/internal/fs"
-	"uncloud/internal/machine/store"
+	"github.com/psviderski/uncloud/internal/machine/store"
+	"github.com/psviderski/uncloud/pkg/api"
 )
 
 const (
@@ -18,7 +19,7 @@ const (
 )
 
 // Controller monitors container changes in the cluster store and generates a configuration file for Caddy reverse
-// proxy. The generated Caddyfile allows Caddy to route external traffic to service containers across the internal
+// proxy. The generated configuration allows Caddy to route external traffic to service containers across the internal
 // network.
 type Controller struct {
 	store          *store.Store
@@ -89,15 +90,20 @@ func (c *Controller) Run(ctx context.Context) error {
 // filterAvailableContainers filters out containers from this machine that are likely unavailable. The availability
 // is determined by the cluster membership state of the machine that the container is running on.
 // TODO: implement machine membership check using Corrossion Admin client.
-func (c *Controller) filterAvailableContainers(containerRecords []store.ContainerRecord) ([]api.Container, error) {
+func (c *Controller) filterAvailableContainers(
-	containers := make([]api.Container, len(containerRecords))
+	containerRecords []store.ContainerRecord,
+) ([]api.ServiceContainer, error) {
+	containers := make([]api.ServiceContainer, len(containerRecords))
 	for i, cr := range containerRecords {
-		containers[i] = cr.Container
+		containers[i] = api.ServiceContainer{
+			Container: cr.Container,
+			// TODO: restore ServiceSpec from the container record once it's saved in the store.
+		}
 	}
 	return containers, nil
 }
 
-func (c *Controller) generateConfig(containers []api.Container) error {
+func (c *Controller) generateConfig(containers []api.ServiceContainer) error {
 	config, err := GenerateConfig(containers, c.verifyResponse)
 	if err != nil {
 		return err

internal/machine/cluster/cluster.go

@@ -11,11 +11,11 @@ import (
 	"log/slog"
 	"net/netip"
 	"time"
-	"uncloud/internal/corrosion"
+	"github.com/psviderski/uncloud/internal/corrosion"
-	"uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
-	"uncloud/internal/machine/network"
+	"github.com/psviderski/uncloud/internal/machine/network"
-	"uncloud/internal/machine/store"
+	"github.com/psviderski/uncloud/internal/machine/store"
-	"uncloud/internal/secret"
+	"github.com/psviderski/uncloud/internal/secret"
 )
 
 type Cluster struct {

internal/machine/cluster/dns.go

@@ -4,12 +4,12 @@ import (
 	"context"
 	"encoding/json"
 	"errors"
+	"github.com/psviderski/uncloud/internal/dns"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
+	"github.com/psviderski/uncloud/internal/machine/store"
 	"google.golang.org/grpc/codes"
 	"google.golang.org/grpc/status"
 	"google.golang.org/protobuf/types/known/emptypb"
-	"uncloud/internal/dns"
-	"uncloud/internal/machine/api/pb"
-	"uncloud/internal/machine/store"
 )
 
 // uncloudDNSKey is the key used to store the details of the reserved domain in the store.

internal/machine/cluster/machine.go

@@ -2,7 +2,7 @@ package cluster
 
 import (
 	"fmt"
-	"uncloud/internal/secret"
+	"github.com/psviderski/uncloud/internal/secret"
 )
 
 // NewMachineID generates a new unique machine ID.

internal/machine/corroservice/config.go

@@ -7,7 +7,7 @@ import (
 	"net/netip"
 	"os"
 	"path/filepath"
-	"uncloud/internal/fs"
+	"github.com/psviderski/uncloud/internal/fs"
 )
 
 const (

internal/machine/db.go

@@ -0,0 +1,51 @@
+package machine
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/jmoiron/sqlx"
+	_ "modernc.org/sqlite"
+)
+
+const DBFileName = "machine.db"
+
+// NewDB creates a new connection to machine SQLite database and runs schema migrations if necessary.
+func NewDB(path string) (*sqlx.DB, error) {
+	// Create the database file with 0600 permissions if it doesn't exist, or update permissions if exists.
+	if _, err := os.Stat(path); os.IsNotExist(err) {
+		file, err := os.OpenFile(path, os.O_CREATE|os.O_RDWR, 0600)
+		if err != nil {
+			return nil, fmt.Errorf("create SQLite database '%s': %w", path, err)
+		}
+		file.Close()
+	} else {
+		if err = os.Chmod(path, 0600); err != nil {
+			return nil, fmt.Errorf("update SQLite database permissions '%s': %w", path, err)
+		}
+	}
+
+	// - Write-Ahead Logging (WAL) mode for better read/write performance.
+	// - Busy timeout (5s) to make concurrent writes wait on each other instead of failing immediately.
+	conn := path + "?_pragma=journal_mode=WAL&_pragma=synchronous=NORMAL&_pragma=busy_timeout=5000&_time_format=sqlite"
+	db, err := sqlx.Connect("sqlite", conn)
+	if err != nil {
+		return nil, fmt.Errorf("connect to SQLite database '%s': %w", conn, err)
+	}
+
+	schema := `
+	CREATE TABLE IF NOT EXISTS containers (
+	    id TEXT NOT NULL PRIMARY KEY,
+	    service_spec TEXT NOT NULL CHECK (json_valid(service_spec)),
+	    -- 'subsecond' modifier is used to store timestamps with millisecond precision.
+	    created_at TIMESTAMP NOT NULL DEFAULT (datetime('subsecond')),
+	    updated_at TIMESTAMP NOT NULL DEFAULT (datetime('subsecond'))
+	);
+    `
+
+	if _, err = db.Exec(schema); err != nil {
+		return nil, fmt.Errorf("create schema: %w", err)
+	}
+
+	return db, nil
+}

internal/machine/dns/resolver.go

@@ -0,0 +1,126 @@
+package dns
+
+import (
+	"context"
+	"fmt"
+	"log/slog"
+	"net/netip"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/psviderski/uncloud/internal/machine/store"
+	"github.com/psviderski/uncloud/pkg/api"
+)
+
+// ClusterResolver implements Resolver by tracking containers in the cluster and resolving service names
+// to their IP addresses.
+type ClusterResolver struct {
+	store *store.Store
+	// serviceIPs maps service names to container IPs.
+	serviceIPs map[string][]netip.Addr
+	// mu protects the serviceIPs map.
+	mu sync.RWMutex
+	// lastUpdate tracks when records were last updated.
+	lastUpdate time.Time
+	log        *slog.Logger
+}
+
+// NewClusterResolver creates a new cluster resolver using the cluster store.
+func NewClusterResolver(store *store.Store) *ClusterResolver {
+	return &ClusterResolver{
+		store:      store,
+		serviceIPs: make(map[string][]netip.Addr),
+		log:        slog.With("component", "dns-resolver"),
+	}
+}
+
+// Run starts watching for container changes and updates DNS records accordingly.
+func (r *ClusterResolver) Run(ctx context.Context) error {
+	containers, changes, err := r.store.SubscribeContainers(ctx)
+	if err != nil {
+		return fmt.Errorf("subscribe to container changes: %w", err)
+	}
+	r.log.Info("Subscribed to container changes in the cluster to keep DNS records updated.")
+
+	// TODO: implement machine membership check using Corrossion Admin client to filter available containers.
+	r.updateServiceIPs(containers)
+
+	for {
+		select {
+		case _, ok := <-changes:
+			if !ok {
+				return fmt.Errorf("containers subscription failed")
+			}
+			r.log.Debug("Cluster containers changed, updating DNS records.")
+
+			containers, err = r.store.ListContainers(ctx, store.ListOptions{})
+			if err != nil {
+				r.log.Error("Failed to list containers.", "err", err)
+				continue
+			}
+
+			// TODO: implement machine membership check using Corrossion Admin client to filter available containers.
+			r.updateServiceIPs(containers)
+		case <-ctx.Done():
+			return nil
+		}
+	}
+}
+
+// updateServiceIPs processes container records and updates the serviceIPs map.
+func (r *ClusterResolver) updateServiceIPs(containers []store.ContainerRecord) {
+	newServiceIPs := make(map[string][]netip.Addr, len(r.serviceIPs))
+
+	containersCount := 0
+	for _, record := range containers {
+		if !record.Container.Healthy() {
+			continue
+		}
+
+		ip := record.Container.UncloudNetworkIP()
+		if !ip.IsValid() {
+			// Container is not connected to the uncloud Docker network (could be host network).
+			continue
+		}
+
+		ctr := api.ServiceContainer{Container: record.Container}
+		if ctr.ServiceID() == "" || ctr.ServiceName() == "" {
+			// Container is not part of a service, skip it.
+			continue
+		}
+
+		// TODO: remove normalisation after implementing service name validation:
+		//.https://github.com/psviderski/uncloud/issues/53
+		serviceName := strings.ToLower(ctr.ServiceName())
+
+		newServiceIPs[serviceName] = append(newServiceIPs[serviceName], ip)
+		// Also add the service ID as a valid lookup.
+		newServiceIPs[ctr.ServiceID()] = append(newServiceIPs[ctr.ServiceID()], ip)
+		containersCount++
+	}
+
+	// Update the serviceIPs map atomically.
+	r.mu.Lock()
+	r.serviceIPs = newServiceIPs
+	r.mu.Unlock()
+
+	r.log.Debug("DNS records updated.", "services", len(newServiceIPs)/2, "containers", containersCount)
+}
+
+// Resolve returns IP addresses of the service containers.
+func (r *ClusterResolver) Resolve(serviceName string) []netip.Addr {
+	r.mu.RLock()
+	defer r.mu.RUnlock()
+
+	ips, ok := r.serviceIPs[serviceName]
+	if !ok || len(ips) == 0 {
+		return nil
+	}
+
+	// Return a copy of the IPs slice to prevent modification of the original.
+	ipsCopy := make([]netip.Addr, len(ips))
+	copy(ipsCopy, ips)
+
+	return ipsCopy
+}

internal/machine/dns/server.go

@@ -0,0 +1,347 @@
+package dns
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"log/slog"
+	"math/rand/v2"
+	"net"
+	"net/netip"
+	"strconv"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/miekg/dns"
+)
+
+const (
+	// InternalDomain is the cluster internal domain for service discovery. All DNS queries ending with this suffix
+	// will be resolved using the internal DNS server.
+	InternalDomain = "internal."
+	// Port is the standard DNS port.
+	Port = 53
+	// maxConcurrentForwards is the maximum number of concurrent forwarded queries to upstream DNS servers.
+	// 1024 is the default used by the Docker internal DNS server.
+	maxConcurrentForwards = 1024
+	// forwardingTimeout is the timeout for forwarding a DNS query to an upstream server.
+	forwardingTimeout = 3 * time.Second
+)
+
+// Resolver is an interface for resolving service names to IP addresses.
+type Resolver interface {
+	// Resolve returns a list of IP addresses of the service containers.
+	// An empty list is returned if no service is found.
+	Resolve(serviceName string) []netip.Addr
+}
+
+// Server is an embedded internal DNS server for service discovery and forwarding external queries
+// to upstream DNS servers.
+type Server struct {
+	listenAddr      netip.Addr
+	resolver        Resolver
+	upstreamServers []netip.AddrPort
+
+	udpServer        *dns.Server
+	tcpServer        *dns.Server
+	inProgressReqs   sync.WaitGroup
+	forwardSemaphore chan struct{}
+	log              *slog.Logger
+}
+
+// NewServer creates a new DNS server with the given configuration.
+// If upstreams is nil, nameservers from /etc/resolv.conf will be used. An empty upstreams list means to only resolve
+// internal DNS queries and not forward any external queries.
+func NewServer(listenAddr netip.Addr, resolver Resolver, upstreams []netip.AddrPort) (*Server, error) {
+	if !listenAddr.IsValid() {
+		return nil, fmt.Errorf("invalid listen address: %s", listenAddr)
+	}
+	if resolver == nil {
+		return nil, fmt.Errorf("resolver must be provided")
+	}
+
+	// Load upstreams from /etc/resolv.conf and set fallback servers only if upstreams is not provided.
+	if upstreams == nil {
+		if nameservers, err := parseNameserversFromResolvConf(); err != nil {
+			slog.Warn("Failed to parse nameservers from /etc/resolv.conf.", "err", err)
+		} else {
+			for _, ns := range nameservers {
+				if ns.Compare(listenAddr) == 0 {
+					// Skip this internal DNS server address if it has been configured in /etc/resolv.conf
+					// to not create a forwarding loop.
+					continue
+				}
+				upstreams = append(upstreams, netip.AddrPortFrom(ns, Port))
+			}
+		}
+
+		// Fallback to common public DNS servers if no nameservers were found in /etc/resolv.conf.
+		if upstreams == nil {
+			upstreams = []netip.AddrPort{
+				netip.AddrPortFrom(netip.MustParseAddr("1.1.1.1"), Port), // Cloudflare DNS
+				netip.AddrPortFrom(netip.MustParseAddr("8.8.8.8"), Port), // Google DNS
+			}
+		}
+	}
+
+	return &Server{
+		listenAddr:       listenAddr,
+		resolver:         resolver,
+		upstreamServers:  upstreams,
+		forwardSemaphore: make(chan struct{}, maxConcurrentForwards),
+		log:              slog.With("component", "dns-server"),
+	}, nil
+}
+
+// ListenAddr returns the address the DNS server is listening on.
+func (s *Server) ListenAddr() netip.Addr {
+	return s.listenAddr
+}
+
+// Run starts the DNS server listening on both UDP and TCP ports. The server on TCP is not critical so it won't return
+// an error if it fails to start. The server will run until the context is canceled or an error occurs.
+func (s *Server) Run(ctx context.Context) error {
+	addr := net.JoinHostPort(s.listenAddr.String(), strconv.Itoa(Port))
+	s.udpServer = &dns.Server{
+		Addr:    addr,
+		Net:     "udp",
+		Handler: dns.HandlerFunc(s.handleRequest),
+	}
+	s.tcpServer = &dns.Server{
+		Addr:    addr,
+		Net:     "tcp",
+		Handler: dns.HandlerFunc(s.handleRequest),
+	}
+
+	errCh := make(chan error, 1) // Buffer size 1 is for UDP server error only.
+
+	go func() {
+		s.log.Info("Starting DNS server on UDP port.", "addr", addr, "upstreams", s.upstreamServers)
+		if err := s.udpServer.ListenAndServe(); err != nil {
+			errCh <- fmt.Errorf("listen and serve on %s/udp: %w", addr, err)
+		}
+	}()
+
+	go func() {
+		s.log.Info("Starting DNS server on TCP port.", "addr", addr, "upstreams", s.upstreamServers)
+		if err := s.tcpServer.ListenAndServe(); err != nil {
+			// TCP server is not critical, so log the error and continue.
+			slog.Warn("Failed to listen and serve DNS server on TCP port. "+
+				"TCP server is not critical and will be ignored.", "addr", addr, "err", err)
+		}
+	}()
+
+	select {
+	case err := <-errCh:
+		// Stop the servers if the UDP one fails.
+		s.stop()
+		return err
+	case <-ctx.Done():
+		s.log.Info("Stopping DNS server.")
+		return s.stop()
+	}
+}
+
+// stop gracefully shuts down the DNS server.
+func (s *Server) stop() error {
+	var udpErr, tcpErr error
+
+	if s.udpServer != nil {
+		udpErr = s.udpServer.Shutdown()
+	}
+	if s.tcpServer != nil {
+		tcpErr = s.tcpServer.Shutdown()
+	}
+
+	// Wait for all in-progress requests to finish.
+	s.inProgressReqs.Wait()
+
+	if udpErr != nil {
+		return fmt.Errorf("shutdown DNS server (UDP): %w", udpErr)
+	}
+
+	if tcpErr != nil {
+		return fmt.Errorf("shutdown DNS server (TCP): %w", tcpErr)
+	}
+
+	return nil
+}
+
+// handleRequest processes a DNS query and returns an appropriate response.
+func (s *Server) handleRequest(w dns.ResponseWriter, req *dns.Msg) {
+	s.inProgressReqs.Add(1)
+	defer s.inProgressReqs.Done()
+
+	if len(req.Question) == 0 {
+		resp := new(dns.Msg).SetRcode(req, dns.RcodeFormatError)
+		s.reply(w, req, resp)
+		return
+	}
+
+	// While the original DNS RFCs allow multiple questions, in practice it never works. So handle only the first one.
+	q := req.Question[0]
+	log := s.log.With("name", q.Name, "type", dns.TypeToString[q.Qtype])
+	log.Debug("Received DNS query.")
+
+	if !dns.IsSubDomain(InternalDomain, dns.CanonicalName(q.Name)) {
+		log.Debug("Forwarding non-internal DNS query to upstream DNS servers.")
+
+		// Use the same transport for the forwarded request as the original request.
+		resp, err := s.forwardRequest(req, w.LocalAddr().Network())
+		if err != nil {
+			log.Error("Failed to forward DNS query.", "err", err)
+			resp = new(dns.Msg).SetRcode(req, dns.RcodeServerFailure)
+		}
+
+		s.reply(w, req, resp)
+		return
+	}
+
+	// Handle the query for the internal domain.
+	resp := new(dns.Msg).SetReply(req)
+	resp.Authoritative = true
+	resp.RecursionAvailable = true
+
+	switch q.Qtype {
+	case dns.TypeA:
+		records := s.handleAQuery(q.Name)
+		if len(records) > 0 {
+			log.Debug("Found A records for internal DNS query.", "count", len(records))
+			resp.Answer = append(resp.Answer, records...)
+		} else {
+			log.Debug("No records found for internal DNS query.")
+			resp.SetRcode(req, dns.RcodeNameError)
+		}
+		// TODO: Handle other query types (SRV, TXT, etc.) as needed.
+	}
+
+	// Truncate the response if it exceeds the maximum size for the transport protocol.
+	maxSize := dns.MinMsgSize
+	if w.LocalAddr().Network() == "tcp" {
+		maxSize = dns.MaxMsgSize
+	} else {
+		// Retrieve the UDP buffer size from the EDNS0 record if present.
+		if opt := req.IsEdns0(); opt != nil {
+			if udpSize := int(opt.UDPSize()); udpSize > maxSize {
+				maxSize = udpSize
+			}
+		}
+	}
+	resp.Truncate(maxSize)
+
+	s.reply(w, req, resp)
+}
+
+func (s *Server) reply(w dns.ResponseWriter, req *dns.Msg, resp *dns.Msg) error {
+	if err := w.WriteMsg(resp); err != nil {
+		s.log.Error("Failed to write DNS response.", "err", err, "msg", resp)
+		// It may fail due to a malformed response message, e.g. exceeding the maximum size. In that case,
+		// attempt to send a server failure response instead so the client doesn't have to wait for a timeout.
+		if resp.Rcode != dns.RcodeServerFailure {
+			resp = new(dns.Msg).SetRcode(req, dns.RcodeServerFailure)
+			if err2 := w.WriteMsg(resp); err2 != nil {
+				s.log.Error("Failed to write DNS error response.", "err", err2, "msg", resp)
+			}
+		}
+		return err
+	}
+	return nil
+}
+
+// forwardRequest forwards a DNS query to system DNS servers
+func (s *Server) forwardRequest(req *dns.Msg, proto string) (*dns.Msg, error) {
+	if len(s.upstreamServers) == 0 {
+		return nil, errors.New("no upstream DNS servers configured")
+	}
+
+	// Apply concurrency control for forwarded queries.
+	select {
+	case s.forwardSemaphore <- struct{}{}:
+		defer func() {
+			<-s.forwardSemaphore
+		}()
+	default:
+		return nil, fmt.Errorf("too many concurrent forwarded queries (max: %d)", maxConcurrentForwards)
+	}
+
+	// Create DNS client with timeout and same transport as original request
+	client := &dns.Client{
+		Net:     proto,
+		Timeout: forwardingTimeout,
+	}
+
+	var lastErr error
+	for _, server := range s.upstreamServers {
+		resp, _, err := client.Exchange(req, server.String())
+		if err == nil {
+			return resp, nil
+		}
+		lastErr = err
+		s.log.Debug("Failed to forward DNS query to upstream server.", "server", server, "err", err)
+	}
+
+	return nil, lastErr
+}
+
+// handleAQuery processes an A query for the internal domain and returns A records for the requested name.
+// The internal domain suffix is already stripped from the name. An empty list is returned if no records are found.
+func (s *Server) handleAQuery(name string) []dns.RR {
+	serviceName := trimInternalDomain(name)
+	ips := s.resolver.Resolve(serviceName)
+	if len(ips) == 0 {
+		s.log.Debug("Failed to resolve service name.", "service", serviceName)
+		return nil
+	}
+	s.log.Debug("Resolved service name.", "service", serviceName, "ips", ips)
+
+	if len(ips) > 1 {
+		// TODO: sort by proximity to the requesting container/machine. For now, just shuffle the IPs.
+		rand.Shuffle(len(ips), func(i, j int) {
+			ips[i], ips[j] = ips[j], ips[i]
+		})
+	}
+
+	// Create A records for each IP.
+	records := make([]dns.RR, 0, len(ips))
+	for _, ip := range ips {
+		records = append(records, &dns.A{
+			Hdr: dns.RR_Header{
+				Name:   name,
+				Rrtype: dns.TypeA,
+				Class:  dns.ClassINET,
+				// TODO: should we increate the TTL to some reasonably small value like 5-30 seconds to allow
+				//  at least some caching?
+				Ttl: 0,
+			},
+			A: net.ParseIP(ip.String()),
+		})
+	}
+	return records
+}
+
+// parseNameserversFromResolvConf parses the nameservers from /etc/resolv.conf.
+func parseNameserversFromResolvConf() ([]netip.Addr, error) {
+	config, err := dns.ClientConfigFromFile("/etc/resolv.conf")
+	if err != nil {
+		return nil, err
+	}
+
+	var servers []netip.Addr
+	for _, server := range config.Servers {
+		if addr, err := netip.ParseAddr(server); err == nil {
+			servers = append(servers, addr)
+		}
+	}
+
+	return servers, nil
+}
+
+func trimInternalDomain(name string) string {
+	name = dns.CanonicalName(name)
+	if !dns.IsSubDomain(InternalDomain, name) {
+		return name
+	}
+
+	return strings.TrimSuffix(name, "."+InternalDomain)
+}