Permanent condprof TEST-UNEXPECTED-TIMEOUT | automation.py | application timed out after 370 seconds with no output when version increase hasn't been merged to autoland yet
Categories
(Testing :: Performance, defect, P5)
Tracking
(firefox-esr91 unaffected, firefox-esr102 unaffected, firefox102 unaffected, firefox103 disabled, firefox111 fixed)
Tracking | Status | |
---|---|---|
firefox-esr91 | --- | unaffected |
firefox-esr102 | --- | unaffected |
firefox102 | --- | unaffected |
firefox103 | --- | disabled |
firefox111 | --- | fixed |
People
(Reporter: intermittent-bug-filer, Assigned: jmaher)
References
(Regression)
Details
(Keywords: intermittent-failure, regression, Whiteboard: [stockwell unknown])
Attachments
(2 files)
Filed by: archaeopteryx [at] coole-files.de
Parsed log: https://treeherder.mozilla.org/logviewer?job_id=382653011&repo=autoland
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/CTf_roecQL6-SKZSN9GgGQ/runs/0/artifacts/public/logs/live_backing.log
mozilla-central got the version increased from 103 to 104 and a new conditioned profile artifact got created. On autoland, the version is still 103 (merge back is scheduled after the next merge from autoland to mozilla-central, also merging autoland to mozilla-central could be blocked for older, good merge candidates).
The tests with conditioned profile fail on autoland because the application has a lower version number and rejects to launch with the profile with a higher version number.
![]() |
||
Updated•3 years ago
|
Updated•3 years ago
|
Comment hidden (Intermittent Failures Robot) |
Comment 2•3 years ago
|
||
These didn't show up beginning with this range, looks like they're fixed.
Comment 3•3 years ago
|
||
:jmaher, this is very likely to come back again. One idea to resolve this could be to somehow perform a profile upgrade in condprof just before the test. This would be similar to having it upgraded on mozilla-central.
Assignee | ||
Comment 4•3 years ago
|
||
:sparky, I am leaning towards a versioned profile in the index- then we can retrigger older stuff. Do you have thoughts on versioned profiles vs upgrade in place?
Comment 5•3 years ago
|
||
Good idea :jmaher! If you can trigger condprof to run on version changes in autoland, that should resolve this issue (assuming version changes are the only thing that impact the profile compatibility).
Assignee | ||
Comment 6•3 years ago
|
||
oh, I wasn't thinking that- I might consider that. I was thinking of just calling the artifact profile-v104-linux64-full.tgz
or something like that. I guess if there are other needs to upgrade the profile then we are sort of screwed. I have a mostly completed patch to add versions with a fallback to existing profiles- not sure if it makes sense to continue on with that, or investigate ways to upload artifacts to an index from autoland and trigger as needed
Comment 7•3 years ago
|
||
Is there someone we could ask about whether version changes are the only culprit or if there are others?
This is why I suggested upgrading the profile with condprof before using it in the test. I'm not suggesting a full re-run of the conditioning scenarios, but just opening the profile, allowing an upgrade, then using it in the test. It's relatively simple to use condprof within the test harnesses also, here's where we make use of it in raptor: https://searchfox.org/mozilla-central/source/testing/raptor/raptor/perftest.py#274
If you set force_new
to false, then it'll take the latest archive produced on m-c and update it. You'd need to add some code to skip running the scenarios though: https://searchfox.org/mozilla-central/source/testing/condprofile/condprof/creator.py#180
Assignee | ||
Comment 8•3 years ago
|
||
that seems straightforward, launch the profile, let it upgrade, then use it for copies for tests that need it.
in a scenario where we need to downgrade, we are sort of screwed. I think we need to take both approaches:
- add an artifact which includes the version number so we can always reference it. This will allow for "downgrades" or older try pushes to work
- add a profile upgrade step after downloading a profile, basically download, run firefox and let it upgrade, then copy the profile around as needed for tests.
Comment 9•3 years ago
|
||
Good point! I agree, using both approaches solves the issue. The only question now is if the version is enough.
Assignee | ||
Comment 10•3 years ago
|
||
Updated•3 years ago
|
Comment 11•3 years ago
|
||
A patch has been attached on this bug, which was already closed. Filing a separate bug will ensure better tracking. If this was not by mistake and further action is needed, please alert the appropriate party.
Assignee | ||
Updated•3 years ago
|
Comment hidden (Intermittent Failures Robot) |
Comment 13•3 years ago
|
||
Comment 14•3 years ago
|
||
bugherder |
Assignee | ||
Comment 15•3 years ago
|
||
going to fix this to update the INDEX with a version, not just the artifact name.
Updated•3 years ago
|
Comment hidden (Intermittent Failures Robot) |
Updated•3 years ago
|
Comment hidden (Intermittent Failures Robot) |
Comment hidden (Intermittent Failures Robot) |
Comment hidden (Intermittent Failures Robot) |
Comment hidden (Intermittent Failures Robot) |
Comment 22•3 years ago
|
||
There have been 43 total failures in the last 7 days, recent failure log.
Affected platforms are:
- linux1804-64-shippable-qr
- windows10-64-2004-qr
- windows10-64-2004-shippable-qr
[task 2022-09-21T19:00:54.750Z] 19:00:54 INFO - TEST-INFO | started process GECKO(824)
[task 2022-09-21T19:00:55.214Z] 19:00:55 INFO - GECKO(824) | 1663786855226 Marionette INFO Marionette enabled
[task 2022-09-21T19:00:55.228Z] 19:00:55 INFO - GECKO(824) | 1663786855228 Marionette TRACE Received observer notification final-ui-startup
[task 2022-09-21T19:00:55.230Z] 19:00:55 INFO - GECKO(824) | 1663786855229 Marionette INFO Listening on port 2828
[task 2022-09-21T19:00:55.231Z] 19:00:55 INFO - GECKO(824) | 1663786855229 Marionette DEBUG Marionette is listening
[task 2022-09-21T19:03:55.735Z] 19:03:55 INFO - runtests.py | Waiting for browser...
[task 2022-09-21T19:07:05.246Z] 19:07:05 INFO - Buffered messages finished
[task 2022-09-21T19:07:05.248Z] 19:07:05 ERROR - TEST-UNEXPECTED-TIMEOUT | automation.py | application timed out after 370 seconds with no output
[task 2022-09-21T19:07:05.248Z] 19:07:05 ERROR - Force-terminating active process(es).
[task 2022-09-21T19:07:05.249Z] 19:07:05 INFO - Determining child pids from psutil...
[task 2022-09-21T19:07:05.253Z] 19:07:05 INFO - [8264]
[task 2022-09-21T19:07:05.254Z] 19:07:05 INFO - ==> process 8264 launched child process 2360 ("Z:\task_166378584991358\build\application\firefox\firefox.exe" -contentproc --channel="8264.0.787247084\2020129246" -parentBuildID 20220921180701 -prefsHandle 1900 -prefMapHandle 1892 -prefsLen 21845 -prefMapSize 255114 -appDir "Z:\task_166378584991358\build\application\firefox\browser" - {c1a50efb-7386-47b1-ae34-f73b13038234} 8264 "\\.\pipe\gecko-crash-server-pipe.8264" 1980 158a2df8258 gpu)
[task 2022-09-21T19:07:05.254Z] 19:07:05 INFO - Found child pids: {8264, 2360}
[task 2022-09-21T19:07:05.255Z] 19:07:05 INFO - Killing process: 8264
[task 2022-09-21T19:07:05.261Z] 19:07:05 INFO - TEST-INFO | started process screenshot
[task 2022-09-21T19:07:05.335Z] 19:07:05 INFO - TEST-INFO | screenshot: exit 0
Comment hidden (Intermittent Failures Robot) |
Assignee | ||
Comment 24•3 years ago
|
||
another issue with bumping firefox versions- one of these days I need to fix this.
Comment 25•3 years ago
|
||
https://wiki.mozilla.org/Bug_Triage#Intermittent_Test_Failure_Cleanup
For more information, please visit auto_nag documentation.
Comment 26•3 years ago
|
||
This seems to happen again? Plain central run: https://treeherder.mozilla.org/jobs?repo=try&revision=f0a9b60b2c0471af9dd9e385a68bc5070d33cd10
Assignee | ||
Comment 27•3 years ago
|
||
we never did get a fix for condprof tasks and the version change. My simple half day of hacking didn't work. The good news is this seems to be the last remaining condprof issue (we had a lot of small issues when we first turned these on)
Reporter | ||
Comment 28•3 years ago
|
||
treeherder |
New failure instance: https://treeherder.mozilla.org/logviewer?job_id=399360530&repo=autoland
Comment hidden (Intermittent Failures Robot) |
Comment hidden (Intermittent Failures Robot) |
Comment 31•3 years ago
|
||
https://wiki.mozilla.org/Bug_Triage#Intermittent_Test_Failure_Cleanup
For more information, please visit auto_nag documentation.
Reporter | ||
Comment 32•3 years ago
|
||
treeherder |
New failure instance: https://treeherder.mozilla.org/logviewer?job_id=402539489&repo=autoland
Comment 33•3 years ago
|
||
Joel, could you have another go at this? Each release cycle we're hitting this again and it's not really pleasant. Thank you.
Comment hidden (Intermittent Failures Robot) |
Comment hidden (Intermittent Failures Robot) |
Assignee | ||
Comment 36•3 years ago
|
||
Assignee | ||
Comment 37•3 years ago
|
||
we will know in a few weeks if this fix works reliably.
Comment 38•3 years ago
|
||
Comment 39•3 years ago
|
||
bugherder |
Comment 40•2 years ago
|
||
I will reopen this because it is happening again on autoland only for now. On linux/windows https://treeherder.mozilla.org/logviewer?job_id=411863229&repo=autoland / https://treeherder.mozilla.org/logviewer?job_id=411860037&repo=autoland
Assignee | ||
Comment 41•2 years ago
|
||
this failed again, what happened was:
- we bumped firefox to v114.0a1
- condprof client tried to download m-c.latest.firefox v113.0a1
- it fell back to 2 days prior 2023.04.08.firefox.latest
- this failed because it didn't exist- only 1 task on m-c ran that day and it was a fixed_by_commit: https://treeherder.mozilla.org/jobs?repo=mozilla-central&searchStr=firefox-full&revision=80d1079b264cb63cae33dca7270a13dad4269708
- finally went with latest which is firefox v114
so this is an edge case- really the only case I saw in the last 6 weeks.
:sparky- any thoughts on what to do here? leave it alone, fix something?
Comment hidden (Intermittent Failures Robot) |
Comment 43•2 years ago
|
||
I agree, I don't think we'll see this happen too often either so we can leave it alone.
Comment hidden (Intermittent Failures Robot) |
Comment hidden (Intermittent Failures Robot) |
Comment hidden (Intermittent Failures Robot) |
Comment 47•2 years ago
|
||
https://wiki.mozilla.org/Bug_Triage#Intermittent_Test_Failure_Cleanup
For more information, please visit BugBot documentation.
Description
•