autoscaler: fix premature end of binpacking #6165
Merged
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Outdated suggestions cannot be applied.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.
What type of PR is this?
/kind bug
What this PR does / why we need it:
When PermissionToAddNode gets called without actually adding a new node, the node counter in the thresholdBasedEstimationLimiter gets out of sync with the actual number of new nodes. This can happen when the "is the last node empty" check triggers.
The solution used here is to rearrange the checks so that PermissionToAddNode is followed by adding a new node. Alternatively, it might also be possible to pass the current number of nodes as parameter.
Special notes for your reviewer:
Found while investigating scale up scenarios with dynamic resource allocation support, see https://kubernetes.slack.com/archives/C09R1LV8S/p1695991972759879.
Does this PR introduce a user-facing change?