action #39629
closed
openQA Scheduler refactor fallout
Added by szarate over 5 years ago.
Updated over 5 years ago.
Category:
Feature requests
Description
This is going to be a general ticket to track problems with the new scheduler with support for blocked_by deployed during last week
Currently known problems are mostly related to jobs that are ran, when the parent is still not even started
- Description updated (diff)
- Description updated (diff)
- Related to action #32725: [tools] Scheduler job_grab/filter_jobs refactoring added
- Related to action #39560: Tests for blocked_by and loops inside of it added
As a result after having a full build, and seeing jobs, that were missing certain parts:
And many other, with a beta on top, it was decided to revert the changes (at obs level) and deploy them in OSD for the time being. While we look at the blocked_by whole changes a bit better
https://progress.opensuse.org/issues/39560#note-4
Also, stuck in assigned (still in that condition):
- Related to action #39068: Webui killed by out of memory in o3 (triggered by postgresql) added
- Status changed from New to Resolved
We found in the second round several bugs that were fixed and are now 'good enough' in production. We have 2 more issues to be fixed in future sprints though:
- Usability of how cluster scheduling are to debugged by reviewers (#40772)
- Starvation of multimachine jobs (#48011)
https://progress.opensuse.org/issues/40904 needs to be fixed in the spec file
- Target version changed from Current Sprint to Done
Also available in: Atom
PDF