Bug #10083

queued jobs in Paused pipeline instances start running anyway

Added by Joshua Randall over 2 years ago. Updated over 1 year ago.

Status:
New
Priority:
Normal
Assigned To:
-
Category:
-
Target version:
Start date:
09/16/2016
Due date:
% Done:

0%

Estimated time:
Story points:
-

Description

I had a set of ~1000 pipeline instances each of which with a single job. 27 of the pipeline instances had a running job (one for each of our 27 crunch nodes). The rest of the queued pipeline instances were changed to '{"state":"Paused"}' using the CLI.

My expectation was that the 27 jobs that were currently running would complete, but no new jobs would be started. Thus, eventually I'd have 27 free nodes on which to start a task that requires all nodes.

Instead, as jobs completed, new jobs were started.

My guess would be that because the jobs had already been queued, they were not effected by pausing the pipeline instances. I am not sure how I could have achieved what I wanted.

I ended up cancelling all 1000 jobs.

History

#1 Updated by Tom Morris over 1 year ago

  • Target version set to Arvados Future Sprints

Also available in: Atom PDF