Sign in
Agents

Sync Running Runs

Reconcile stale agent runs. Called by Kubernetes CronJob. Internal endpoint, no auth (relies on cluster-internal access).

Two passes:

  1. Re-dispatch stale 'pending' runs — messages lost during worker restart (Dramatiq acks on receive, so restarts lose in-flight messages).
  2. Timeout stale 'running' runs — worker crash without Shutdown handler.
post/api/internal/agents/sync

Response

Successful Response

{"stackTrail":"paths:/api/internal/agents/sync:post:responses:200:content:application/json:schema","oasType":"schema","type":"unknown"}

Changes