Every source has a live status and a streaming log view, so you can see what a pipeline is doing without SSHing into a container or a Pod.
Where to look
- Status — the source detail page shows the runner's state (see lifecycle).
- Live logs — the runner's own stdout/stderr, streamed live with auto-refresh (
docker logson the compose stack, the Pod's log on a cluster — same stream, same view). - Health — runner health rows written at commit cadence, so a transient DB blip doesn't take the status endpoint down.
Common failure modes
| Symptom | What to check |
|---|---|
| Runner won't start | The runner's log (docker logs runner-<connector>, kubectl logs pod/<runner-pod>, or the Live logs panel), the generated config, the Celery worker |
| Schema discovery fails | Database credentials, network reachability, logs/fabric.log |
| UI not loading | nginx running, dist/ files present |
| Source held | The named setting from the preflight / schema-guard refusal |
These are actionable next steps, not just symptoms. When a preflight or schema guard refuses, the refusal names the setting so you can fix it and restart.