Background Jobs
Long-running work runs asynchronously so the UI stays responsive. Track everything under Jobs, where each job carries a human-friendly ID (e.g. EC-1), a phase, and a live log.
| Job type | Triggered by |
|---|---|
| SNAPSHOT_UPDATE | Publishing resource changes to the control-plane. |
| WAF_PROPAGATION | Pushing a saved WAF configuration to proxies. |
| ACME_VERIFICATION | Certificate issuance / DNS-01 validation. |
| RESOURCE_UPGRADE | Migrating resources to a new Envoy version. |
Jobs move through ANALYZING → PENDING → CLAIMED → RUNNING → COMPLETED/FAILED.
Monitoring the job system
The Jobs view surfaces the whole async subsystem, backed by these controller endpoints:
| View | What it shows |
|---|---|
Stats (GET /jobs/stats) | Aggregate counts by phase — how much work is queued, running, and failed. |
Workers (GET /jobs/workers) | The worker-pool status — how many workers are alive and what they're claiming. |
Stuck jobs (GET /jobs/stuck) | Jobs whose heartbeat has gone stale (> 5 min) — a worker died or a step hung. |
Retry (POST /jobs/:id/retry) | Re-run a failed (or stuck) job from the beginning. |
A healthy system shows workers alive, few or no stuck jobs, and failures clearing on retry. Persistent stuck jobs or a failing retry point at an upstream problem (control-plane unreachable, a bad config, a DNS-01 that can't validate) — open the job's live log to see the failing phase.