The "Add New Jobs" is not displaying the list of jobs. Tested on today's nightly and Firefox 48 both with fresh profiles. After selecting "Add New Jobs", nothing appears, but the menu is updated to show "Hide Runnable Jobs".
I think I've seen this before. Can you share the URL of the Treeherder page you're seeing this on?
Since filing the bug, I was able to run new jobs, but most of the time I'm not seeing the list. Unscientific success rate for me is about 1/15. I'm seeing it on the two URL's below. https://treeherder.mozilla.org/#/jobs?repo=try&revision=524f93eb4ce3 https://treeherder.mozilla.org/#/jobs?repo=try&revision=e44136e4e686
Sometimes the request to get runnable jobs is taking a long time to finish (6+ seconds). Other times, I'm seeing occasional 500 error responses to https://treeherder.mozilla.org/api/project/try/runnable_jobs/?decisionTaskID=cpfPyKtjSFO63D22oT2Lrg With the response being an object containing single character properties that combine to form the string "Exception: 404 Client Error: Not Found for url: https://queue.taskcluster.net/v1/task/cpfPyKtjSFO63D22oT2Lrg/artifacts/public/full-task-graph.json" Without seeing the logs for these requests on the server, I don't know what is happening to cause this. Wildly flailing with the needinfos to find someone who can look into this further. :)
There aren't any useful errors on the web nodes, other than a bunch of read errors between apache and gunicorn, and gunicorn worker timeouts. NewRelic shows a pair of huge spikes in log_parser transaction times right around then (bug 1296434), but I'm not sure if that would be related. It looks like something wonky on the TaskCluster side, though?
2 years ago
In bug 1288028 I mention how we could have an up-to-date runnable jobs table at all times, thus, not needing to fetch an artifact by the API.
Summary: "Add New Jobs" doesn't show the list of jobs to add → If fetching runnable jobs fails we don't notify the user that "Add New Jobs" failed to work
Comment on attachment 8832811 [details] [review] [treeherder] KWierso:1297185 > mozilla:master This adds some functions to catch failure cases on the two promises called while fetching the list of runnable jobs. In the failure case, it pops up an error notification, but it still leaves the push in the "add new jobs" mode (so the menu option shows as "hide runnable jobs"). Not sure if I should make the error messages persistent or let them disappear after a few seconds. Also not sure if I should just put the push back into the default mode so that the menu option shows as "add new jobs".
Attachment #8832811 - Flags: review?(cdawson)
I didn't really test it enough when actually hitting the runnable jobs apis to see if the intermittent failures get caught by this PR, but the new error messages show up if I load treeherder, unplug from the network, then press the "add new jobs" button. If you want to test both error cases, you'd need to add a delay inside the first promise's success case so that you can turn off the network before the second promise gets called.
Comment on attachment 8832811 [details] [review] [treeherder] KWierso:1297185 > mozilla:master Looks good to me. Since we know it'll hit the other error notify if an error is hit, I think we're fine here.
Attachment #8832811 - Flags: review?(cdawson) → review+
Commit pushed to master at https://github.com/mozilla/treeherder https://github.com/mozilla/treeherder/commit/64775346e17d5ddbd310cf430269ff1319c320ff Bug 1297185 - Show an error if fetching runnable jobs fails (#2124) r=camd
Status: NEW → RESOLVED
Last Resolved: a year ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.