Repository navigation
CI: capture hang dumps on unit-test jobs - #8653
Merged
Aaronontheweb merged 2 commits intoSep 28, 2026
Merged
Conversation
Akka.Tests has hung ~4 times in ~480 builds on the Linux unit-test job, with silence until Incrementalist force-cancels the command at its 20-minute per-project timeout, discarding the .trx and leaving no clue which test hung. Add --blame-hang --blame-hang-timeout 15m --blame-hang-dump-type mini to both unit-test jobs (Windows and Linux). 15m stays under Incrementalist's 20m timeout, so vstest's Blame collector fires first, kills the hung testhost, and writes a Sequence_*.xml + mini dump into TestResults before Incrementalist would otherwise force-kill with no diagnostics. Those files are already swept into the existing OS-specific job artifact by the shared template's TestResults/** copy, so no template changes are needed. Not applied to the multi-node (MNTR) jobs: they share the dotnet test invocation but run separate node processes, so hang detection there is a different question.
Incrementalist's 20m timeout stops waiting on dotnet test but never kills it, so blame fires on its own clock rather than racing Incrementalist. A full dump is needed to walk async state in an all-threads-idle hang.
Aaronontheweb
enabled auto-merge (squash)
September 28, 2026 18:13
Aaronontheweb
disabled auto-merge
September 28, 2026 18:46
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Akka.Tests has hung on the Linux unit-test job a handful of times (~4 in ~480 builds), and we can't tell which test hung. Incrementalist gives up waiting after its 20-minute per-project timeout, and the job ends with no diagnostics.
This adds
--blame-hang --blame-hang-timeout 15m --blame-hang-dump-type fullto the Windows and Linux.NET Unit Testsjobs inbuild-system/pr-validation.yaml. If one test runs 15 minutes with no test starting or finishing, vstest's Blame collector takes a full dump of the testhost and kills it. It records the hung test inSequence_*.xmland the .trx.dotnet testand never kills it, so blame still fires after that. A hang that starts late in the last project to run can still end with the job, undumped.TestResults/**copy puts these files into each job's OS-specific artifact, so no template changes were needed.This only captures the next hang; it doesn't fix it. It isn't applied to the multi-node (MNTR) jobs, where a hang could be in a separate node process rather than the vstest testhost.