#!watchflakes
default <- pkg == "golang.org/x/tools/gopls/internal/test/marker" && test == "Test/fixedbugs/issue59944.txt"

Issue created automatically to collect these failures.

Example (log):

=== RUN   Test/fixedbugs/issue59944.txt
=== PAUSE Test/fixedbugs/issue59944.txt
=== CONT  Test/fixedbugs/issue59944.txt

— watchflakes

Comment From: gopherbot

Found new dashboard test flakes for:

#!watchflakes
default <- pkg == "golang.org/x/tools/gopls/internal/test/marker" && test == "Test/fixedbugs/issue59944.txt"
2025-06-17 17:42 x_tools-go1.23-openbsd-amd64 tools@03042518 release-branch.go1.23@3eedbde2 x/tools/gopls/internal/test/marker.Test/fixedbugs/issue59944.txt [SKIP] (log) === RUN Test/fixedbugs/issue59944.txt === PAUSE Test/fixedbugs/issue59944.txt === CONT Test/fixedbugs/issue59944.txt

— watchflakes

Comment From: adonovan

The only information in the log is:

FAIL    golang.org/x/tools/gopls/internal/test/marker   1202.284s

How does this happen? Even a timeout should print something more than just FAIL.

Hmm, LUCI adds this interpolation:

go test reported no status for this test. This usually means execution was interrupted by SIGTERM, SIGABORT or SIGKILL.

Oh... it seems that although main LUCI output is very quiet, the JSON log contains a backtrace of thousands of goroutines (search for panic). Strange that this information should be buried like that.

Comment From: gopherbot

Change https://go.dev/cl/682836 mentions this issue: main.star: increase openbsd-amd64 timeout scaling to 4