-
Notifications
You must be signed in to change notification settings - Fork 17.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
x/sys/unix: TestPoll failures with "wrong number of events" beginning 2021-11-04 #49380
Comments
(Marking as release-blocker for Go 1.18 because this appears to be a regression on linux-amd64, which is a first-class port.) |
I'll look into it. I think you're probably right. |
My attempts to reproduce this have been unsuccessful thus far. Is this particularly rare? |
This failure mode is also so strange... It creates a FIFO that it then blocks on, and waits to time out. How could something be sent on that FIFO? What else could possibly get that FD? |
Would it make sense to augment the test to print which events it thinks fired? Even if I can't reproduce locally, it should in theory pop up on the builders eventually. |
It does seem to only occur in a small fraction of runs: we have a ton of (But we've also had ∞% more failures than we did between whenever my If we think it may have been fixed since Nov. 6, maybe we can park it in WaitingForInfo and see if it occurs again by the time the Gopherbot expires it?
2021-11-06T13:10:06-c75c477-4f083c7/linux-amd64-race |
I'm always in favor of more-descriptive test logs. 👍 |
Oops. I forgot to mark it as |
Change https://golang.org/cl/363660 mentions this issue: |
For golang/go#49380 Change-Id: Ie1d370681962d9f69ef54b33ddf38e4c74a2e298 Reviewed-on: https://go-review.googlesource.com/c/sys/+/363660 Trust: Ian Lance Taylor <iant@golang.org> Run-TryBot: Ian Lance Taylor <iant@golang.org> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Michael Knyszek <mknyszek@google.com>
Looks fixed to me.
|
greplogs --dashboard -md -l -e '(?m)FAIL: TestPoll .*\n\s+.*wrong number of events'
2021-11-05T07:00:05-7861aae-6fefb7f/linux-amd64
2021-11-05T00:52:09-7861aae-bd580a0/linux-amd64-race
2021-11-04T21:41:49-7861aae-156abe5/linux-amd64-race
These all occurred after CL 361255 in
x/sys
, but I don't see how it could possibly be related. I suspect a regression on the Go side — @mknyszek, could this be another test that is overly sensitive to being interrupted by a GC cycle?The text was updated successfully, but these errors were encountered: