HTTP edge metrics from Caddy
Per-service latency, error rate, status codes. From the proxy your apps already pass through. No agent in your app. No code changes.
Coming soon
See what's happening
in production
Uncloud Hub is the UI and Grafana stack for your Uncloud clusters.
Get real-time metrics, logs, and alerts out of the box to catch issues early and fix
faster.
Hub is built for Uncloud, the open-source orchestrator that runs your apps across machines.
What you actually see
Pre-built dashboards and alerts for the failure modes you care about.
Logs next to the spike
When something goes red, the relevant application logs are right there. No hunting for log files across machines. No new query language to learn.
Slow Postgres queries
Hub watches every Uncloud-managed Postgres for queries eating your latency budget. Find them before your users do.
Alerts that mean something
Pre-built rules for the failures you actually care about. Service unavailable. Edge errors climbing. Disk filling. Postgres replication lag.
How it works
Connect, collect, see
Three steps. The same stack you'd build by hand, set up for you.
1
Connect
Got an Uncloud cluster running? Connect it in one command. New to Uncloud? Hub walks you through setting one up from your own machines.
$ uc hub connect
2
Install the agent
Hub detects what's running. Caddy as your ingress, Postgres on machine 2, your worker, your web app. It configures the right collectors for each. No exporters to install by hand.
✓ Detected Caddy ingress
✓ Detected Postgres 16
✓ Configuring exporters
3
See
Within minutes, dashboards are live. Failed HTTP requests, slow queries, application logs, system health. Plus alerts that page when users are affected.
4 dashboards live
12 alert rules active
logs streaming
Same Grafana stack you'd have built. Full access.
Hub does the setup and the ongoing maintenance for you. It ships with opinions about what's worth watching. Underneath, it's still your data and your tools.
-
Click any chart, see the queryNo black boxes. The PromQL is right there, ready to copy or edit.
-
Customise our dashboards or add your ownFull Grafana editor. Your panels live alongside the pre-built ones.
-
Hand the data to an AI agentRead-only data API. Ask plain-English questions about your fleet. It's your data.
-
Out of the critical pathIf Hub goes down, your apps keep running. Hub watches. It does not run things.
Built on the back of Uncloud
5K+
GitHub stars
19K+
Downloads
2x
Hacker News front page
In use
Radboud University
A note from us
I'm Pasha. I've spent a decade building commercial software and keeping production infrastructure healthy. Uncloud started because I wanted a simpler way to run my own apps across cloud VMs and machines at home, without a Kubernetes degree.
Hub is the dashboard we needed for that, productised. If the problem above sounds like your weekend, We'd like to hear about your setup. The design partner round is small. The feedback we get now is what shapes what ships.
Where Hub is going
Hub starts with observability because that's what we keep hearing from people running things on their own servers. After that, the direction is broader. Shared cluster access without juggling SSH keys. Connectivity for machines behind NAT. A clean way to reach your cluster from your laptop as if you were already inside it.
None of this is on the page yet because none of it is shipping yet. Observability lands first. The rest follows once the wedge is real.
Frequently asked questions
What's open source and what's paid?
Uncloud itself stays open source under Apache-2.0. The on-cluster agent that connects to Hub will be open source too, with a configurable endpoint. The Hub backend is closed source and paid.
Does it work without Uncloud?
No, by design. Hub knows what's running because Uncloud manages it. That auto-discovery is the difference between Hub and a generic Grafana you set up yourself.
Can my team log in too?
Single user for now. Multi-user and team access are planned. If single user is a blocker for you, please tell us on the form below. That's a strong signal for prioritising it sooner.
Can I self-host Hub itself?
Planned for orgs that need to keep everything on their own infrastructure. Not in the first release.
I already run Grafana. What now?
Keep it if it's working. Hub can sit alongside as a separate dashboard for the Uncloud-aware views. If you'd rather stop maintaining your own Grafana, Hub replaces it entirely. Same Grafana underneath either way.
When does it ship and how much does it cost?
First version is in design now. Pricing is per cluster, designed to land below the cost of the machines you're watching. The form below is the way into the design partner round.
What if Hub is down?
Your apps keep running. Hub watches your cluster from outside it and is not in the critical path of your traffic, your data, or your deploys. If hub.uncloud.run becomes unavailable, you lose dashboards and alerts for that period. You do not lose your applications.
Help shape Hub
We're picking 5 to 10 design partners for the first version. If you run things on a few servers and the Day 2 part is what's grinding you down, We'd like to hear about your setup.
Or drop your email below and we'll send updates as Hub takes shape.