This task is a placeholder for my suspicion: unfortunately I don't have any evidence handy but recently I have seen that a build was scheduled in the buildbot but not actually run afterwards. This maybe might something to do with the fact that we don't have enough builders (#331) and pending tasks are thrown away if they don't get clients in time? I don't find it too likely though. According to what I saw this happens to tasks that are scheduled while other high time-consuming task is running (depcheck). Will provide more info if available or close if all of this is false. :)
So I figured out what the problem is. Look at https://taskotron.stg.fedoraproject.org/taskmaster/builders/x86_64/builds/46375. Changes are grouped together and only one build is triggered for all of them and not one build per each change. Solving this might also fix #330.
@tflink I think setting treeStableTimer to None in JobScheduler might fix it. Let me know if you can think of better way, otherwise I'll try that on dev. Thanks.
treeStableTimer
None
JobScheduler
@mkrizek - that sounds good to me, making the change on dev sounds like a good way to test
Possible fix deployed to dev, will keep an eye on it if it works or not.
I think setting treeStableTimer to None in JobScheduler might fix it
In addition I passed mergeRequests=False to builders and it seems the problem is fixed, I rather wait until tomorrow to make sure though.
mergeRequests=False
Deployed to dev and stg. According to my testing, this seems to be fixed. We can alwasys reopen if we see that happen again.
Metadata Update from @mkrizek: - Issue tagged with: infrastructure
Log in to comment on this ticket.