fix(shim): synchronously check cgroup for OOM at container exit#12843
Open
kpurdon wants to merge 1 commit intogoogle:masterfrom
Open
fix(shim): synchronously check cgroup for OOM at container exit#12843kpurdon wants to merge 1 commit intogoogle:masterfrom
kpurdon wants to merge 1 commit intogoogle:masterfrom
Conversation
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
kpurdon
force-pushed
the
kpurdon/12838/fix-aarch64-oom-reporting
branch
2 times, most recently
from
da73340 to
89555b6
Compare
On aarch64, the async OOM notification via inotify/EventChan (cgroups v2) loses the race against the container exit notification from runtime.Wait(). This causes containerd to process the TaskExit before TaskOOM arrives, leading kubelet to report reason=Error instead of reason=OOMKilled. Add a synchronous check of the cgroup's memory.events at container exit time. When OOM kills are detected, publish the TaskOOM event through the same sequential event pipeline as TaskExit, guaranteeing the OOM event arrives first. The async and sync paths coordinate through a shared lastOOM map under mutex to prevent duplicate events. Fixes google#12838
kpurdon
force-pushed
the
kpurdon/12838/fix-aarch64-oom-reporting
branch
from
89555b6 to
f1c7618
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
On aarch64 (ARM64/Graviton), the async OOM notification via inotify/
EventChan(cgroups v2) loses the race against the container exit fromruntime.Wait(). Containerd processes theTaskExitbeforeTaskOOMarrives, so kubelet reportsreason: Errorinstead ofreason: OOMKilled. This breaks monitoring, autoscaling, and alerting on aarch64 gVisor nodes.What
isOOM(id string) boolto theoomPollerinterface for synchronous OOM checking at exit time (one-shot: consumes the cgroup reference)isOOMfor cgroups v2 (watcherV2) usingcg.Stat()to readmemory.eventsdirectly, bypassing the async inotify pathlastOOMMapfrom goroutine-local inrun()to a shared mutex-protected field, coordinating between async and sync paths to prevent duplicateTaskOOMeventsisOOMfor cgroups v1 (epoller) since its eventfd mechanism doesn't have this raceFixes #12838