cmd/go: TestScript/version_buildvcs_git_gpg (if enabled) fails on linux longtest builders #57034
Comments
bcmills
added
Testing
|
@gopherbot, please backport to Go 1.19 and 1.18: this test may fail spuriously due to problems with a fragile third-party tool. |
|
Backport issue(s) opened: #57054 (for 1.18), #57055 (for 1.19). Remember to create the cherry-pick CL(s) as soon as the patch is submitted to master, according to https://go.dev/wiki/MinorReleases. |
|
Change https://go.dev/cl/454502 mentions this issue: |
|
Change https://go.dev/cl/454955 mentions this issue: |
|
Change https://go.dev/cl/454956 mentions this issue: |
|
I will update the |
|
@cagedmantis, I don't think you need to bother. I don't plan to try to restore the test — the complexity of the workarounds for the fragile socket behavior does not seem worth it for the incremental value of the test. |
|
@bcmills I will remove |
|
Change https://go.dev/cl/455860 mentions this issue: |
The tests which require gpg and gpg-agent have been removed. This change removes them from the container images. For golang/go#57034 Change-Id: Id33dbc9881bb9a688955612a66f1f549822f3866 Reviewed-on: https://go-review.googlesource.com/c/build/+/455860 Reviewed-by: Bryan Mills <bcmills@google.com> Reviewed-by: Cherry Mui <cherryyz@google.com> Run-TryBot: Carlos Amedee <carlos@golang.org> TryBot-Result: Gopher Robot <gobot@golang.org> Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
…t_gpg This was a regression test added for a 'git' command line used for build stamping. Unfortunately, 'gpg' has proved to be extremely fragile: * In recent versions, it appears to always require 'gpg-agent' to be installed for anything involving secret keys, but for some reason is not normally marked as requiring gpg-agent in Debian's package manager. * It tries to create a Unix domain socket in a subdirectory of $TMPDIR without checking the path length, which fails when $TMPDIR is too long to fit in the 'sun_path' field of a sockaddr_un struct (which typically tops out somewhere between 92 and 108 bytes). We could theoretically address those by artificially reducing the script's TMPDIR length and checking for gpg-agent in addition to gpg, but arguably those should both be fixed upstream instead. On balance, the incremental value that this test provides does not seem worth the complexity of dealing with such a fragile third-party tool. Updates #50675. Updates #48802. Updates #57034. Fixes #57054. Change-Id: Ia3288c2f84f8db86ddfa139b4d1c0112d67079ef Reviewed-on: https://go-review.googlesource.com/c/go/+/454502 TryBot-Result: Gopher Robot <gobot@golang.org> Run-TryBot: Bryan Mills <bcmills@google.com> Auto-Submit: Bryan Mills <bcmills@google.com> Reviewed-by: Cherry Mui <cherryyz@google.com> (cherry picked from commit 45f5ef4) Reviewed-on: https://go-review.googlesource.com/c/go/+/454956
…t_gpg This was a regression test added for a 'git' command line used for build stamping. Unfortunately, 'gpg' has proved to be extremely fragile: * In recent versions, it appears to always require 'gpg-agent' to be installed for anything involving secret keys, but for some reason is not normally marked as requiring gpg-agent in Debian's package manager. * It tries to create a Unix domain socket in a subdirectory of $TMPDIR without checking the path length, which fails when $TMPDIR is too long to fit in the 'sun_path' field of a sockaddr_un struct (which typically tops out somewhere between 92 and 108 bytes). We could theoretically address those by artificially reducing the script's TMPDIR length and checking for gpg-agent in addition to gpg, but arguably those should both be fixed upstream instead. On balance, the incremental value that this test provides does not seem worth the complexity of dealing with such a fragile third-party tool. Updates #50675. Updates #48802. Updates #57034. Fixes #57055. Change-Id: Ia3288c2f84f8db86ddfa139b4d1c0112d67079ef Reviewed-on: https://go-review.googlesource.com/c/go/+/454502 TryBot-Result: Gopher Robot <gobot@golang.org> Run-TryBot: Bryan Mills <bcmills@google.com> Auto-Submit: Bryan Mills <bcmills@google.com> Reviewed-by: Cherry Mui <cherryyz@google.com> (cherry picked from commit 45f5ef4) Reviewed-on: https://go-review.googlesource.com/c/go/+/454955
Split from #49649 . Filing a new issue to investigate/decide.
After setting up the linux-arm64-longtest builder, we saw cmd/go's TestScript/version_buildvcs_git_gpg fails.
https://build.golang.org/log/7b3bdfcc706b20aeb4deb0a6b7fb6c73d75670d9
On 386 and amd64, linux-386-longtest and linux-amd64-longtest builders didn't actually have
gpgcommand installed, so the test was actually skipped. @cagedmantis 's CL installs the command. And after that, the test fails on those builders, too. https://build.golang.org/log/30e41a5b422d7ca512aa244816e1ee578dfa9285So we currently rolled back the builder update.
Interestingly, the test passes under SSH, but fails in the post-submit run and under
gomote run. From #49649 (comment), it seemsTMPDIRor some other environment variable settings may cause the difference.cc @bcmills
The text was updated successfully, but these errors were encountered: