Increase timeout multiplier for opt and debug CCOV builds (4)
Categories
(Testing :: web-platform-tests, task)
Tracking
(Not tracked)
People
(Reporter: whimboo, Unassigned)
References
Details
Attachments
(1 file)
Currently the timeout multiplier for CCOV jobs are set to 2 which is too less. We should increase the value to probably 4. This will not fix all the issues that we currently see on bug 1783217 but at least work towards lesser failure reports. Restarting Firefox a lot within a test module will still remain an issue, which is not perfectly covered by this timeout.
|
Reporter | |
Comment 1•3 years ago
|
||
|
||
Comment 4•3 years ago
|
||
| bugherder | ||
|
|
Reporter | |
Comment 6•3 years ago
|
||
With the known regressions and no easy way to fix the underlying issue in wptrunner I would like to request a backout of this patch.
|
||
Comment 7•3 years ago
|
||
Backed out as requested: https://hg.mozilla.org/integration/autoland/rev/6b15a563499a5350d4dc7407eb6163c3f390aaa4
|
|
Reporter | |
Comment 8•3 years ago
|
||
(In reply to James Graham [:jgraham] from bug 1774933 comment #15)
Yeah, this is unfortunate. We don't propogate the timeout multiplier to the server, so things like trickle don't know about it. That sounds wrong, and is obviously wrong in this case, but it can also be what you want if you're trying to allow for a slower browser (e.g. if you had the opposite situation to these tests, and wanted to check that something happened after a network request you wouldn't want to slow down the request).
Anyway, I think these tests are broken, but it's somewhat difficult to fix them without having a way to pass the timeout multiplier into the delay. Which we don't have at the moment. We should file an upstream issue on that.
I've filed https://github.com/web-platform-tests/wpt/issues/36094 for that.
The broken tests here were:
- /html/semantics/scripting-1/the-script-element/moving-between-documents/ordering/delay-load-event-1.html
- /websockets/keeping-connection-open/001.html?wpt_flags=h2
|
|
Reporter | |
Comment 9•3 years ago
|
||
I'm un-assigning myself for now until it's clear what the next steps are.
|
|
||
Comment 10•3 years ago
|
||
Backout merged to central: https://hg.mozilla.org/mozilla-central/rev/6b15a563499a
|
|
Reporter | |
Comment 12•3 years ago
|
||
(In reply to Web Platform Test Sync Bot (Matrix: #interop:mozilla.org) from comment #11)
Created web-platform-tests PR
https://github.com/web-platform-tests/wpt/pull/36180 for changes under
testing/web-platform/tests
Note that the backout was not merged upstream due to failures. I've closed/reopened the PR to re-run the tests.
|
||
Comment 15•2 years ago
|
||
There's a r+ patch which didn't land and no activity in this bug for 2 weeks.
:whimboo, could you have a look please?
If you still have some work to do, you can add an action "Plan Changes" in Phabricator.
For more information, please visit auto_nag documentation.
|
|
Reporter | |
Comment 16•2 years ago
|
||
(In reply to Release mgmt bot [:suhaib / :marco/ :calixte] from comment #15)
There's a r+ patch which didn't land and no activity in this bug for 2 weeks.
:whimboo, could you have a look please?
This is basically waiting on https://github.com/web-platform-tests/wpt/issues/36094. It would be great if James could have a look at this issue. As such removing my own needinfo.
|
||
Updated•2 years ago
|
|
|
Reporter | |
Comment 17•2 years ago
|
||
I just tried again given that some other changes to the multiplier got landed a while ago but the tests are still failing:
https://treeherder.mozilla.org/jobs?repo=try&revision=fd20b80ff9dbe102357671454e3813a631c9b444
Now I wonder if we should only change the multiplier for CCOV builds when the test_type is wdspec for now.
James, would you be ok with that? If yes I will file a new bug and keep that one open for all other test types.
|
|
||
Comment 18•2 years ago
|
||
I think doing this for just wdspec tests is more likely to work out well, so no objections to trying that.
|
|
Reporter | |
Comment 19•2 years ago
|
||
Thanks. Filed bug 1810785 for increasing the timeout for wdspec.
|
|
Reporter | |
Comment 20•2 years ago
|
||
Given that we had problems with landing this change for all the wpt jobs, I'm going to only modify the timeout multiplier for CCOV builds for wdspec jobs only via bug 1848121. That is fine for us.
Description
•