Closed Bug 1394989 Opened 2 years ago Closed 2 years ago
Permafailing CCov TEST-UNEXPECTED-TIMEOUT | xpcom/tests/unit/test
_ns IProcess _stress .js | Test timed out
Filed by: wkocher [at] mozilla.com https://treeherder.mozilla.org/logviewer.html#?job_id=124836863&repo=mozilla-central https://queue.taskcluster.net/v1/task/YwtMnbjWTMOjL_RPHOGHdQ/runs/0/artifacts/public/logs/live_backing.log This started with this merge from inbound to m-c: https://treeherder.mozilla.org/#/jobs?repo=mozilla-central&revision=eb72c8c077518c3fdc0dca5c0a14070bd25fe3cb&filter-searchStr=7958c8d0ce4c0d2c0ce8da5cb5ae5ad74b02e993
Summary: Intermittent CCov TEST-UNEXPECTED-TIMEOUT | xpcom/tests/unit/test_nsIProcess_stress.js | Test timed out → Permafailing CCov TEST-UNEXPECTED-TIMEOUT | xpcom/tests/unit/test_nsIProcess_stress.js | Test timed out
this failure is quite frequent, we should address this in the short term: https://brasstacks.mozilla.com/orangefactor/index.html?display=Bug&bugid=1394989 (35 failures!) :gmierz, do you have some cycles for this? If not, I can look elsewhere
Whiteboard: [stockwell needswork]
I think I'll have more time to look at this next week so you can get someone else on it if you'd like or I'll try to work on it then. I looked into it a bit and the test started failing between these pushes: https://treeherder.mozilla.org/#/jobs?repo=mozilla-central&filter-searchStr=ccov%2Fopt-xpcshell-8&fromchange=db2a49c83d53&tochange=eb72c8c07751 It also fails here: https://dxr.mozilla.org/mozilla-central/source/dom/media/CubebUtils.cpp#367 But the log says it fails at line 321 of that file. The test itself is creating and killing 1000 processes so I wonder if the code coverage process is interfering with it.
does it even make sense to run ccov on this test? :froydnj, do you have thoughts on this test and what it covers? why it might be failing under code coverage? who might be able to help?
(In reply to Joel Maher ( :jmaher) (UTC-5) from comment #3) > does it even make sense to run ccov on this test? Probably not; it's code that's not really executed during normal browser usage. > :froydnj, do you have thoughts on this test and what it covers? why it > might be failing under code coverage? who might be able to help? I have no idea why it's failing under code coverage. Maybe forking is more expensive with code coverage enabled?
thanks for the reply :froydnj! I see we do not run this test on windows, do we need to run this on osx as well? I am always looking for ways to reduce load there, only if needed. this patch avoids the test on coverage, I am happy to keep it as is or update it as needed :)
Attachment #8903269 - Flags: review?(nfroyd)
Attachment #8903269 - Flags: review?(nfroyd) → review+
Pushed by email@example.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/d57464f03409 Disable xpcom/tests/unit/test_nsIProcess_stress.js for coverage builds. r=froydnj
I am going to let this get resolved/fixed as there is nothing specific to investigate here.
Whiteboard: [stockwell needswork] → [stockwell disabled]
This still runs and permafails: https://treeherder.mozilla.org/logviewer.html#?job_id=128123821&repo=mozilla-central
"coverage" doesn't seem to be set for CCov, at least osfile's test_telemetry.js also runs but should be disabled: https://hg.mozilla.org/mozilla-central/annotate/f5d99e9f3d7bf19c8a222db97bedc924da27fd58/toolkit/components/osfile/tests/xpcshell/xpcshell.ini#l https://public-artifacts.taskcluster.net/Xd-Y1gpkQXKW7ErMhI18NQ/0/public/logs/live_backing.log
This is expected because the 'coverage' flag is for linux64-jsdcov. For linux64-ccov, it's 'ccov' . I've posted a patch for review that fixes this. We've got a bug on file to change the jsdcov flag from 'coverage' to 'jsdcov' in the near future.
Comment on attachment 8904075 [details] Bug 1394989 - Prevent test_nsIProcess.js from running on linux64-ccov. Thank you for the quick fix.
Attachment #8904075 - Flags: review?(aryx.bugmail) → review+
Pushed by firstname.lastname@example.org: https://hg.mozilla.org/integration/mozilla-inbound/rev/8b224e79a2f3 Prevent test_nsIProcess.js from running on linux64-ccov. r=Aryx
Whiteboard: [stockwell disabled]
You need to log in before you can comment on or make changes to this bug.