Repository navigation
diagnostics: ignore zombies when waiting for the process group - #645
Merged
ankitjoju-google merged 1 commit intoNov 10, 2026
Conversation
After a timeout kill, groupAlive polled kill(-pgid, 0) for up to 5s. Without a reaping init (a container whose PID 1 never waits), the killed children stay zombies and still count as group members, so every timed-out collection waited the full 5s and marked the executor busy. Treat a group whose only members are zombies as gone.
jkru3
approved these changes
Nov 10, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
After a collection times out, the collector SIGKILLs the script's process group and waits for it to exit.
groupAliveonly checkedkill(-pgid, 0), which still succeeds while zombie members remain. Without a reaping init (a container whose PID 1 never waits, e.g. Prow's test pod), the killed children never get reaped, so every timed-out run waited the full 5s and setbusy.groupAlivenow also scans/proc/<pid>/statand treats a group whose only members are zombies as gone (it still reports alive if/proccan't be read).GKE nodes aren't affected today because the device plugin pod runs with
hostPID, so the host reaps orphans.Testing:
go test ./pkg/gpu/nvidia/diagnostics/TestCollectorTimeoutKillsProcessGroupwith the test binary as a non-reaping PID 1 (unshare -Urpf --mount-proc): 5.5s and failing before, 0.5s and passing after.