fix: deduplicate discovered machine IPs when adding a new machine

cmd/uncloud/caddy/deploy.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"fmt"
 	"github.com/charmbracelet/huh"
+	"github.com/docker/cli/cli/streams"
 	"github.com/docker/compose/v2/pkg/progress"
 	"github.com/spf13/cobra"
 	"maps"
@@ -173,7 +174,11 @@ func deploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
 	}
 
 	fmt.Println()
-	if _, err = clusterClient.GetDomain(ctx); err != nil {
+	return UpdateDomainRecords(ctx, clusterClient, uncli.ProgressOut())
+}
+
+func UpdateDomainRecords(ctx context.Context, clusterClient *client.Client, progressOut *streams.Out) error {
+	if _, err := clusterClient.GetDomain(ctx); err != nil {
 		if errors.Is(err, client.ErrNotFound) {
 			fmt.Println("Skipping DNS records update as no cluster domain is reserved (see 'uc dns').")
 			return nil
@@ -185,10 +190,11 @@ func deploy(ctx context.Context, uncli *cli.CLI, opts deployOptions) error {
 	// 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
-	}, uncli.ProgressOut(), "Verifying internet access to caddy service")
+	}, progressOut, "Verifying internet access to caddy service")
 	if err != nil {
 		if errors.Is(err, client.ErrNoReachableMachines) {
 			fmt.Println()

cmd/uncloud/machine/add.go

@@ -12,6 +12,7 @@ import (
 	"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"
@@ -138,12 +139,18 @@ func add(ctx context.Context, uncli *cli.CLI, remoteMachine cli.RemoteMachine, o
 		return fmt.Errorf("create caddy deployment: %w", err)
 	}
 
-	return progress.RunWithTitle(ctx, func(ctx context.Context) error {
+	err = progress.RunWithTitle(ctx, func(ctx context.Context) error {
 		if _, err = d.Run(ctx); err != nil {
 			return fmt.Errorf("deploy caddy: %w", err)
 		}
 		return nil
 	}, uncli.ProgressOut(), fmt.Sprintf("Deploying service %s", d.Spec.Name))
+	if err != nil {
+		return err
+	}
+
+	fmt.Println()
+	return caddy.UpdateDomainRecords(ctx, machineClient, uncli.ProgressOut())
 }
 
 func waitClusterInitialised(ctx context.Context, client *client.Client) error {

cmd/uncloud/machine/init.go

@@ -6,15 +6,20 @@ import (
 	"github.com/docker/compose/v2/pkg/progress"
 	"github.com/spf13/cobra"
 	"net/netip"
+	"uncloud/cmd/uncloud/caddy"
+	"uncloud/cmd/uncloud/dns"
 	"uncloud/internal/cli"
 	"uncloud/internal/cli/config"
+	"uncloud/internal/machine/api/pb"
 	"uncloud/internal/machine/cluster"
 )
 
 type initOptions struct {
+	dnsEndpoint string
 	name        string
 	network     string
 	noCaddy     bool
+	noDNS       bool
 	publicIP    string
 	sshKey      string
 	cluster     string
@@ -47,6 +52,8 @@ func NewInitCommand() *cobra.Command {
 			return initCluster(cmd.Context(), uncli, remoteMachine, opts)
 		},
 	}
+	cmd.Flags().StringVar(&opts.dnsEndpoint, "dns-endpoint", dns.DefaultUncloudDNSAPIEndpoint,
+		"API endpoint for the Uncloud DNS service.")
 	cmd.Flags().StringVarP(
 		&opts.name, "name", "n", "",
 		"Assign a name to the machine.",
@@ -59,6 +66,10 @@ func NewInitCommand() *cobra.Command {
 		&opts.noCaddy, "no-caddy", false,
 		"Don't deploy Caddy reverse proxy service to the machine.",
 	)
+	cmd.Flags().BoolVar(
+		&opts.noDNS, "no-dns", false,
+		"Don't reserve a cluster domain in Uncloud DNS.",
+	)
 	cmd.Flags().StringVar(
 		&opts.publicIP, "public-ip", "auto",
 		"Public IP address of the machine for ingress configuration. Use 'auto' for automatic detection, "+
@@ -102,7 +113,7 @@ func initCluster(ctx context.Context, uncli *cli.CLI, remoteMachine *cli.RemoteM
 	}
 	defer client.Close()
 
-	if opts.noCaddy {
+	if opts.noCaddy && opts.noDNS {
 		return nil
 	}
 
@@ -111,15 +122,33 @@ func initCluster(ctx context.Context, uncli *cli.CLI, remoteMachine *cli.RemoteM
 	// after cluster initialisation, we keep the user informed during this wait.
 	fmt.Println("Waiting for the machine to be ready...")
 
+	if !opts.noDNS {
+		domain, err := client.ReserveDomain(ctx, &pb.ReserveDomainRequest{Endpoint: opts.dnsEndpoint})
+		if err != nil {
+			return fmt.Errorf("reserve cluster domain in Uncloud DNS: %w", err)
+		}
+		fmt.Printf("Reserved cluster domain: %s\n", domain.Name)
+	}
+
+	if !opts.noCaddy {
 		d, err := client.NewCaddyDeployment("", nil)
 		if err != nil {
 			return fmt.Errorf("create caddy deployment: %w", err)
 		}
 
-	return progress.RunWithTitle(ctx, func(ctx context.Context) error {
+		err = progress.RunWithTitle(ctx, func(ctx context.Context) error {
 			if _, err = d.Run(ctx); err != nil {
 				return fmt.Errorf("deploy caddy: %w", err)
 			}
 			return nil
 		}, uncli.ProgressOut(), fmt.Sprintf("Deploying service %s", d.Spec.Name))
+		if err != nil {
+			return err
+		}
+
+		fmt.Println()
+		return caddy.UpdateDomainRecords(ctx, client, uncli.ProgressOut())
+	}
+
+	return nil
 }

internal/cli/client/caddy.go

@@ -31,7 +31,6 @@ func (cli *Client) NewCaddyDeployment(image string, filter MachineFilter) (*Depl
 		image = reference.FamiliarString(latest)
 	}
 
-	// TODO: set restart policy to always. https://github.com/psviderski/uncloud/issues/26
 	spec := api.ServiceSpec{
 		Container: api.ContainerSpec{
 			Command: []string{"caddy", "run", "-c", "/config/caddy.json", "--watch"},

internal/machine/machine.go

@@ -19,6 +19,7 @@ import (
 	"os"
 	"os/user"
 	"path/filepath"
+	"slices"
 	"strconv"
 	"uncloud/internal/corrosion"
 	"uncloud/internal/docker"
@@ -540,7 +541,7 @@ func (m *Machine) InitCluster(ctx context.Context, req *pb.InitClusterRequest) (
 	}
 	publicIP, pubIPErr := network.GetPublicIP()
 	// Ignore the error if failed to get the public IP using API services.
-	if pubIPErr == nil {
+	if pubIPErr == nil && !slices.Contains(ips, publicIP) {
 		ips = append(ips, publicIP)
 	}
 	endpoints := make([]*pb.IPPort, len(ips))
@@ -682,7 +683,7 @@ func (m *Machine) Token(_ context.Context, _ *emptypb.Empty) (*pb.TokenResponse,
 	}
 	publicIP, err := network.GetPublicIP()
 	// Ignore the error if failed to get the public IP using API services.
-	if err == nil {
+	if err == nil && !slices.Contains(ips, publicIP) {
 		ips = append(ips, publicIP)
 	}
 	endpoints := make([]netip.AddrPort, len(ips))

scripts/install.sh

@@ -47,7 +47,7 @@ Uncloud supports only systemd-based Linux systems for now."
 }
 
 install_docker() {
-    if command_exists docker; then
+    if command_exists dockerd; then
         log "✓ Docker is already installed."
         docker version
         return
@@ -111,23 +111,37 @@ install_uncloud_binaries() {
     trap "rm -rf '$tmp_dir'" EXIT
 
     local uncloudd_url
+    local uninstall_url
     if [ "${UNCLOUD_VERSION}" == "latest" ]; then
         uncloudd_url="${UNCLOUD_GITHUB_URL}/releases/latest/download/uncloudd_linux_${file_arch}.tar.gz"
+        uninstall_url="https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/uninstall.sh"
     else
         uncloudd_url="${UNCLOUD_GITHUB_URL}/releases/download/${UNCLOUD_VERSION}/uncloudd_linux_${file_arch}.tar.gz"
+        uninstall_url="https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/${UNCLOUD_VERSION}/scripts/uninstall.sh"
     fi
     local uncloudd_download_path="${tmp_dir}/uncloudd.tar.gz"
+    local uninstall_download_path="${tmp_dir}/uninstall.sh"
 
     log "⏳ Downloading uncloudd binary: ${uncloudd_url}"
     if ! curl -fsSL -o "${uncloudd_download_path}" "${uncloudd_url}"; then
         error "Failed to download uncloudd binary."
     fi
-    tar -xf "${uncloudd_download_path}"
-    if ! install ./uncloudd "${uncloudd_install_path}"; then
+    tar -xf "${uncloudd_download_path}" --directory "${tmp_dir}"
+    if ! install "${tmp_dir}/uncloudd" "${uncloudd_install_path}"; then
         error "Failed to install uncloud binary to ${uncloudd_install_path}"
     fi
     log "✓ uncloudd binary installed: ${uncloudd_install_path}"
 
+    log "⏳ Downloading uninstall script: ${uninstall_url}"
+    if ! curl -fsSL -o "${uninstall_download_path}" "${uninstall_url}"; then
+        error "Failed to download uninstall script."
+    fi
+    local uninstall_install_path="${INSTALL_BIN_DIR}/uncloud-uninstall"
+    if ! install "${uninstall_download_path}" "${uninstall_install_path}"; then
+        error "Failed to install uninstall.sh script to ${uninstall_install_path}"
+    fi
+    log "✓ uncloud-uninstall script installed: ${uninstall_install_path}"
+
     # TODO: install uncloud CLI binary and create a uc alias.
 }
 

scripts/uninstall.sh

@@ -0,0 +1,138 @@
+#!/usr/bin/env bash
+
+set -euo pipefail
+
+AUTO_CONFIRM=${AUTO_CONFIRM:-false}
+# Define variables based on the same defaults used in the installation script.
+INSTALL_BIN_DIR=${INSTALL_BIN_DIR:-/usr/local/bin}
+INSTALL_SYSTEMD_DIR=${INSTALL_SYSTEMD_DIR:-/etc/systemd/system}
+UNCLOUD_USER="uncloud"
+UNCLOUD_DATA_DIR=${UNCLOUD_DATA_DIR:-/var/lib/uncloud}
+UNCLOUD_RUN_DIR=${UNCLOUD_RUN_DIR:-/run/uncloud}
+
+log() {
+    echo -e "\033[1;32m$1\033[0m"
+}
+
+error() {
+    echo -e "\033[1;31mERROR: $1\033[0m" >&2
+    exit 1
+}
+
+confirm() {
+    if [ "${AUTO_CONFIRM}" = "true" ]; then
+        return 0
+    fi
+
+    read -r -p "$1 [y/N] " response
+    case "$response" in
+        [yY][eE][sS]|[yY])
+            return 0
+            ;;
+        *)
+            return 1
+            ;;
+    esac
+}
+
+# Check if user is root.
+if [ "$EUID" -ne 0 ]; then
+    error "Please run the uninstall script with sudo or as root."
+fi
+
+# Confirm before proceeding.
+log "⚠️This script will uninstall Uncloud and remove ALL Uncloud managed containers on this machine."
+log "The following actions will be performed:"
+echo "- Remove Uncloud systemd services"
+echo "- Remove Uncloud binaries and data"
+echo "- Remove Uncloud user and group"
+echo "- Remove all Docker containers managed by Uncloud"
+echo "- Remove Uncloud Docker network"
+echo "- Remove Uncloud WireGuard interface"
+
+if ! confirm "Do you want to proceed with uninstallation?"; then
+    log "Uninstallation cancelled."
+    exit 0
+fi
+
+log "⏳ Stopping systemd services..."
+systemctl stop uncloud.service || log "uncloud.service not running or doesn't exist."
+systemctl stop uncloud-corrosion.service || log "uncloud-corrosion.service not running or doesn't exist."
+systemctl disable uncloud.service || log "uncloud.service already disabled or doesn't exist."
+systemctl disable uncloud-corrosion.service || log "uncloud-corrosion.service already disabled or doesn't exist."
+log "✓ Systemd services stopped."
+
+log "⏳ Removing systemd service files..."
+rm -fv "${INSTALL_SYSTEMD_DIR}/uncloud.service"
+rm -fv "${INSTALL_SYSTEMD_DIR}/uncloud-corrosion.service"
+systemctl daemon-reload
+log "✓ Systemd service files removed."
+
+log "⏳ Removing binaries..."
+rm -fv "${INSTALL_BIN_DIR}/uncloudd"
+rm -fv "${INSTALL_BIN_DIR}/uncloud-corrosion"
+log "✓ Binaries removed."
+
+log "⏳ Removing data and run directories..."
+rm -rfv "${UNCLOUD_DATA_DIR}"
+rm -rfv "${UNCLOUD_RUN_DIR}"
+log "✓ Data and run directories removed."
+
+log "⏳ Removing Linux user and group..."
+if id "${UNCLOUD_USER}" &> /dev/null; then
+    userdel "${UNCLOUD_USER}" || error "Failed to remove user ${UNCLOUD_USER}, it may still be used by other processes."
+    log "✓ Linux user '${UNCLOUD_USER}' removed."
+else
+    log "Linux user '${UNCLOUD_USER}' does not exist."
+fi
+
+# Check if the group still exists (other users may be using it).
+if getent group "${UNCLOUD_USER}" &> /dev/null; then
+    groupdel "${UNCLOUD_USER}" || error "Failed to remove group ${UNCLOUD_USER}, it may still be used by other processes or users."
+    log "✓ Linux group '${UNCLOUD_USER}' removed."
+else
+    log "Linux group '${UNCLOUD_USER}' does not exist or was already removed."
+fi
+
+log "⏳ Looking for Docker containers and network created by Uncloud..."
+if command -v docker &> /dev/null; then
+    uncloud_containers=$(docker ps -a --filter "label=uncloud.managed" -q)
+    if [ -n "${uncloud_containers}" ]; then
+        log "Found $(echo "${uncloud_containers}" | wc -w) Uncloud managed containers."
+        log "⏳ Stopping Uncloud managed containers..."
+        echo "${uncloud_containers}" | xargs docker stop || error "Failed to stop Uncloud managed containers."
+        log "⏳ Removing Uncloud managed containers..."
+        echo "${uncloud_containers}" | xargs docker rm || error "Failed to remove Uncloud managed containers."
+        log "✓ Uncloud managed containers stopped and removed."
+    else
+        log "No Uncloud managed containers found."
+    fi
+
+    uncloud_network=$(docker network ls --filter "name=uncloud" -q)
+    if [ -n "${uncloud_network}" ]; then
+        log "⏳ Removing Docker network uncloud..."
+        docker network rm uncloud || error "Failed to remove Docker network uncloud."
+        log "✓ Docker network uncloud removed."
+    else
+        log "Docker network uncloud not found."
+    fi
+else
+    log "Docker CLI not found, skipping Docker container and network cleanup."
+fi
+
+log "⏳ Removing WireGuard interface uncloud..."
+if ip link show uncloud &> /dev/null; then
+    ip link delete uncloud
+    log "✓ WireGuard interface uncloud removed."
+else
+    log "WireGuard interface uncloud not found."
+fi
+
+log "⏳ Removing uninstall script..."
+rm -fv "${INSTALL_BIN_DIR}/uncloud-uninstall"
+log "✓ Uninstall script removed."
+
+echo
+log "✅ Uncloud has been uninstalled successfully!"
+log "Note: Docker installation was preserved. If you want to completely remove Docker as well, \
+follow https://docs.docker.com/engine/install/ubuntu/#uninstall-docker-engine"