[API] Jobs that sit in the front of the queue for a long time without running should be cancelled
If a job sits at the front of the queue without running for some period of time, it should be automatically cancelled on the assumption that it is probably unrunnable. The timeout should be configurable, maybe 24 hours.
#1 Updated by Peter Amstutz about 5 years ago
These jobs have been in the queue for two weeks at the time of this writing:
Trying to cancel the jobs results in fiddlesticks:
#4 Updated by Peter Amstutz almost 5 years ago
- Subject changed from [API] Job has been queued for two weeks and can't be cancelled to [API] Jobs that sit in the front of the queue for a long time without running should be cancelled
- Description updated (diff)
- Target version changed from Arvados Future Sprints to Bug Triage
The fiddlesticks error is covered in #4273. This ticket is now specifically about jobs sitting in the queue for a long time and never running.