-
Notifications
You must be signed in to change notification settings - Fork 0
Closed
Description
Why
Need a reliable worker that executes deferred work, retries on failure, and updates job state. Enables fetching and extraction to run outside the request path
Description of Done
- Single process worker polls the jobs table and executes due jobs
- Concurrency is configurable
- Exponential backoff with jitter on failure
- Job state transitions are persisted with last error text
- Graceful shutdown drains in-flight work
- Unit tests cover backoff, max attempts, and state updates
Tasks
- Define job kinds and dispatch traits
- Implement polling loop with visibility timeout to prevent double work
- Implement exponential backoff with jitter and maximum attempts
- Add cooperative cancellation and shutdown signal handling
- Persist transitions: queued -> running -> succeeded or failed
- Add metric counters and timers via tracing events
- Write unit tests for scheduler, backoff and state transitions
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
enhancementNew feature or requestNew feature or request