While debugging some buildbot bridge stuff I found this task: https://tools.taskcluster.net/task-inspector/#KD5lEi6qSO-PCOhg7exRTg/ In its definition I see: "deadline": "2015-09-14T23:50:28.653Z", "expires": "2016-09-14T23:50:28.653Z", Yet its only run was not resolved as deadline-exceed until 2015-09-15T11:26:08.354Z. I'm probably misunderstanding something here...but it looks like this Task was not expired correctly?
I think it could be our deadline resolution queue that was overloaded, or the deadline resolver had too many tasks to process... I'll take a look at this... IMO: it's not really a critical thing. And the only promise from taskcluster-queue perspective is that unresolved tasks will be resolved deadline-exceeded at some point after task.deadline. Ie. eventually, normally it should be fairly fast, but 11 hours later is still eventually to me :)
OK, if this is expected I guess we have to cope with it...it's quite surprising to not be able to cancel a job whose state is "running" though...
We really should investigate this and see if it's a common problem. IMO, it's just a delay, not anything critical... When you get 409 from a task it's no longer yours to deal with. Anyways, I'm clearing the NI, maybe I'll get around to looking at this some day.
I don't think it should be possible to call: claimTask after task.deadline, but maybe I need to add a test case for this.
Summary: Task not expired correctly? → Investigate/test claimTask after task.deadline
Commits pushed to master at https://github.com/taskcluster/taskcluster-queue https://github.com/taskcluster/taskcluster-queue/commit/15823ada2e01ec8d542243a0323bcf20081056f9 Fix bug 1204898 https://github.com/taskcluster/taskcluster-queue/commit/5ada924b5a674ed4669aed77ec9600d139b5e4a2 Merge pull request #115 from taskcluster/bug1204898 Fix bug 1204898
Status: NEW → RESOLVED
Last Resolved: 2 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.