Machine States & Troubleshooting
The machine card on your Host dashboard always shows a status badge. Most of the time it reads Online (or a progress state like Initializing, Updating GPU Driver, Starting Up). This page covers the states that need your attention — and the ones that just need patience.
Problem states
Section titled “Problem states”| You might see | It means | Do |
|---|---|---|
| Offline | The machine stopped responding entirely — power, network, or VPN is down | Check power and internet. The agent reconnects on its own once the machine is back; nothing to re-register |
| Node Not Ready | The machine is reachable, but its workload runtime isn’t accepting pods | Usually transient and self-healing. If it persists, reboot the machine; still stuck → contact support |
| Agent Not Ready | The Meshive agent on the machine stopped reporting | A reboot restarts the agent. If it keeps happening, reach out on Discord |
| Port Check Failed | One of the four required ports isn’t reachable from outside | Fix the firewall/router rule shown on the card (port/protocol not reachable), then hit the retry button — the card shows Re-checking ports… while it verifies |
| Boot Failed | The machine came up, but a startup step failed | Open the card for the failing step’s message; hardware/cabling issues are the usual cause |
| Error Message on a setup step | A specific onboarding/setup stage failed | The step’s own message says what to fix; a retry button re-runs setup from there |
Offline and not-ready states also show how long ago the machine was last seen, so you can tell a blip from an outage.
States that resolve themselves
Section titled “States that resolve themselves”| Badge | What’s happening |
|---|---|
| Re-verifying | A hardware change or abnormal boot was detected (the badge’s tooltip names the reason — GPU replaced, motherboard changed, thermal shutdown, …) and the mini stress test is re-validating the machine |
| Updating Package / GPU Driver / Agent | Platform-managed updates (how they’re scheduled) |
| Maintenance | Your own maintenance window is active — end or extend it from the card |
| Grace Period (Removal Pending) | The machine is being removed while clients still have workloads on it (lifecycle) |
Email alerts
Section titled “Email alerts”You don’t have to watch the dashboard. Meshive emails you when your machine needs attention — including when the platform detects that the machine can no longer start client workloads or a system component on it has failed. If you get one of those, start with the table above; the email names the failing part.