avalara / node-jobs Goto Github PK

Repro Steps

1. Create job foo
2. Start a worker to process foo, and don't invoke the callback
3. Kill the Worker

Check that the foo job is still marked in the running queue.

We need a way to detect when we have zero workers, and check that the running queue has items in it. For each item left in the item queue, check the job manifest, and if it's not complete, recycle the job by putting it back into the pending queue.

Right now, job manifests (redis hash with key = job.id), are kept indefinitely.

These should expire after some period of time, so they are removed.