make sure MetricAsyncRecorder flush buffer ASAP to avoid overflow #134976
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Outdated suggestions cannot be applied.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.
/kind bug
What this PR does / why we need it:
When a plugin is doing filter, each node will generate a event and send it to
r.bufferCh. meanwhile, the buffer is flushed in every seconds. So if cluster is larger than the buffer size of channel,r.bufferChcan be easily full and overflowed. it cause plugin in later position barely have chance to be seen.This fix make sure the buffer can be flushed ASAP, to avoid channel overflow as much as possible.
Which issue(s) this PR is related to:
Fixes: #134972
Special notes for your reviewer:
Without this fix, the chart in the bottom left, is evaluated by:
The line in the chart look very discontinuous, and in most time the duration of Filter only cost 0.
And with this fix, the chart looks normal now, and no performance regressed:

Furthermore, the data in the chart appears similar to the previous E2E test with 100 nodes: kubernetes/perf-tests#3368
https://prow.k8s.io/?job=ci-kubernetes-e2e-gce-100-node-dra-with-workload
Does this PR introduce a user-facing change?
Additional documentation e.g., KEPs (Kubernetes Enhancement Proposals), usage docs, etc.: