A restart is only useful if you verify what happened afterwards. The right follow-up to a restart is a status check, an RPC probe, and a quick dashboard or log confirmation.
Quick answer
Use openclaw gateway restart, then run openclaw gateway status --require-rpc and openclaw dashboard --no-open to confirm the service is really back.
Command line steps
1. Restart the managed service
Use the gateway command instead of killing processes by hand.
BASH
$openclaw gateway restart
2. Require a passing status probe
If the RPC probe fails, treat that as a failed restart until proven otherwise.
BASH
$openclaw gateway status --require-rpc
openclaw gateway health
3. Confirm the control surface
A fresh dashboard URL is a good final sanity check after the restart.
BASH
$openclaw dashboard --no-open
openclaw logs --limit 200
What to check if it still fails
- If restart succeeds but the RPC probe fails, inspect auth and binding before you touch models or channels.
- If the service is managed by launchd or systemd and keeps failing, run
openclaw doctor --deep. - If the dashboard link works only locally, use SSH forwarding or a tailnet for remote access instead of changing the bind blindly.