x/tools/gopls/internal/regtest: TestGCDetails is flaky on darwin #44099
Comments
gopherbot
added
Tools
findleyr
changed the title
findleyr
changed the title
|
@bcmills Yes, that is certainly possible. Darwin in particular has, in our experience, highly variable timing (we think due to kernel locking). It's odd that I couldn't reproduce on gomote though, given the frequency of flakes -- perhaps I just got lucky. For now, let me double the timeout for this one test case to see if the flakes reduce. |
|
Change https://golang.org/cl/289690 mentions this issue: |
|
Change https://golang.org/cl/289692 mentions this issue: |
Experiment with increasing the timeout for TestGCDetails, given how its flakiness correlates with slow builders. For golang/go#44099 Change-Id: I27a8732256a77c3d9ce2a1b1f75ce5d0fbc11e67 Reviewed-on: https://go-review.googlesource.com/c/tools/+/289690 Trust: Robert Findley <[email protected]> Run-TryBot: Robert Findley <[email protected]> gopls-CI: kokoro <[email protected]> TryBot-Result: Go Bot <[email protected]> Reviewed-by: Rebecca Stambler <[email protected]>
The gopls regtests have highly variable performance on certain operating systems, we believe due to kernel contention. Flakes are a problem, and it's better to have coverage of our default execution mode than no coverage at all. For golang/go#44099 For golang/go#42789 Change-Id: Ie9f3f93bb950930401aac3ce55cdffb889d0f0e1 Reviewed-on: https://go-review.googlesource.com/c/tools/+/289692 Trust: Robert Findley <[email protected]> Run-TryBot: Robert Findley <[email protected]> gopls-CI: kokoro <[email protected]> TryBot-Result: Go Bot <[email protected]> Reviewed-by: Rebecca Stambler <[email protected]>
|
Change https://golang.org/cl/291232 mentions this issue: |
|
Change https://golang.org/cl/291232 mentions this issue: |
This test is flaking at a relatively high rate on darwin and freebsd. Skipping it is better than having a build dashboard with many known failures. Perhaps we could limit this skip to certain GOOS, but for now let's skip it entirely until we understand the problem better. For golang/go#44099 Change-Id: I58703b2db0e5768f75758080d07f9d29b8b5d661 Reviewed-on: https://go-review.googlesource.com/c/tools/+/291232 Trust: Robert Findley <[email protected]> Run-TryBot: Robert Findley <[email protected]> gopls-CI: kokoro <[email protected]> TryBot-Result: Go Bot <[email protected]> Reviewed-by: Rebecca Stambler <[email protected]>
|
Change https://golang.org/cl/304169 mentions this issue: |
At long last, with Pontus's help reproing on Github actions, we have tracked down the race to the gc_details diagnostics. Since toggling gc_details does not increment the snapshot ID, we have to use careful locking to ensure that the gc_details diagnostics we store are consistent with the current state of the gc_details toggle. Updates golang/go#44099 Fixes golang/go#44826 Change-Id: I7b9108a829c98a84360c9012c1b60f4990839b5a Reviewed-on: https://go-review.googlesource.com/c/tools/+/304169 Trust: Robert Findley <[email protected]> Run-TryBot: Robert Findley <[email protected]> gopls-CI: kokoro <[email protected]> TryBot-Result: Go Bot <[email protected]> Reviewed-by: Heschi Kreinick <[email protected]>
|
Change https://golang.org/cl/305409 mentions this issue: |
My hope is that CL 304169 resolved this flake, but we'll have to see. For golang/go#44099 Change-Id: Iac82dc24167daacd0361eac1a5567048a7ecf11c Reviewed-on: https://go-review.googlesource.com/c/tools/+/305409 Trust: Robert Findley <[email protected]> Run-TryBot: Robert Findley <[email protected]> gopls-CI: kokoro <[email protected]> TryBot-Result: Go Bot <[email protected]> Reviewed-by: Heschi Kreinick <[email protected]>
|
This is no longer flaking. |
Recent example:
https://build.golang.org/log/840ef7221cf0147181be16e21a553166c3a9c546
(or click any of the many red buttons in the darwin columns at https://build.golang.org/?repo=golang.org%2fx%2ftools#short :( )
This feature is os-dependent, so it would not be surprising if this is a real bug. Unfortunately, I ran it on gomote and was not able to repro after dozens of attempts. @stamblerre also ran it locally on darwin 1000x and was not able to repro.
CC @heschik
The text was updated successfully, but these errors were encountered: