crypto/tls: high rate of test timeouts on 2022-08-08 #54606
Labels
FrozenDueToAge
NeedsInvestigation
Someone must examine and confirm this is a valid issue and not a duplicate of an existing one.
Milestone
Comments
bcmills
added
the
NeedsInvestigation
bcmills
changed the title
|
One of my CL causes higher failed rate was reverted https://go-review.googlesource.com/c/go/+/422035 |
|
Nice, thank you for the reference! That does seem to explain these failures. I think I spotted the bug, and left a comment about it on https://go.dev/cl/405674. If you'd like to retry that change, you're welcome to mail to to me — I can also suggest some unit-tests that might expose that race. 🙂 |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
greplogs -l -e 'goroutine \d+ \[IO wait, .*\]:\n(?:.+\n\t.+\n)*crypto/tls\.TestServerHandshakeContextCancellation'2022-08-08T15:33:45-0581d69/freebsd-amd64-13_0
2022-08-08T15:33:45-0581d69/openbsd-386-68
2022-08-08T15:22:02-dd59088/dragonfly-amd64-622
2022-08-08T14:12:21-e761556/windows-amd64-2012
2022-08-08T14:12:03-fe406c8/linux-386
2022-08-08T14:12:03-fe406c8/linux-amd64-race
2022-08-08T14:10:37-5405df0/freebsd-386-12_3
2022-08-08T14:10:19-964f0c7/windows-amd64-2012
Since these failures were all clustered around a few CLs on 2022-08-08, it's not clear to me whether this was a regression that has been fixed or a flaky test that happened to correlate in time due to some external factor.
@FiloSottile, @golang/security: are you aware of any changes that would explain this apparent regression and/or its fix?
The text was updated successfully, but these errors were encountered: