[SPARK-49385][K8S] Fix getReusablePVCs to use podCreationTimeout instead of podAllocationDelay
#47867
Conversation
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
…instead of `podAllocationDelay`
HyukjinKwon
approved these changes
Aug 26, 2024
Member
Author
|
Thank you, @HyukjinKwon . Merged to master/3.5/3.4. |
dongjoon-hyun
added a commit
that referenced
this pull request
Aug 26, 2024
…instead of `podAllocationDelay` This PR aims to use `podCreationTimeout` instead of `podAllocationDelay` when `getReusablePVCs` excludes the newly created PVCs of previous batches. K8s control plane pod creation can be delayed due to the unknown reasons. So, `podAllocationDelay (default: 1s)` is insufficient to say that the previous allocation batch's pods are created with their PVCs. We had better wait until `podCreationTimeout`. This affects only the initial set of executors because the baseline is PVC's `getCreationTimestamp`. So, this fixes only a buggy situation where a PVC is shared by two executors due to the long pending executor pod. Pass the CIs with newly updated test cases. No. Closes #47867 from dongjoon-hyun/SPARK-49385. Authored-by: Dongjoon Hyun <dongjoon@apache.org> Signed-off-by: Dongjoon Hyun <dongjoon@apache.org> (cherry picked from commit f596079) Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>
dongjoon-hyun
added a commit
that referenced
this pull request
Aug 26, 2024
…instead of `podAllocationDelay` This PR aims to use `podCreationTimeout` instead of `podAllocationDelay` when `getReusablePVCs` excludes the newly created PVCs of previous batches. K8s control plane pod creation can be delayed due to the unknown reasons. So, `podAllocationDelay (default: 1s)` is insufficient to say that the previous allocation batch's pods are created with their PVCs. We had better wait until `podCreationTimeout`. This affects only the initial set of executors because the baseline is PVC's `getCreationTimestamp`. So, this fixes only a buggy situation where a PVC is shared by two executors due to the long pending executor pod. Pass the CIs with newly updated test cases. No. Closes #47867 from dongjoon-hyun/SPARK-49385. Authored-by: Dongjoon Hyun <dongjoon@apache.org> Signed-off-by: Dongjoon Hyun <dongjoon@apache.org> (cherry picked from commit f596079) Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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 changes were proposed in this pull request?
This PR aims to use
podCreationTimeoutinstead ofpodAllocationDelaywhengetReusablePVCsexcludes the newly created PVCs of previous batches.Why are the changes needed?
K8s control plane pod creation can be delayed due to the unknown reasons. So,
podAllocationDelay (default: 1s)is insufficient to say that the previous allocation batch's pods are created with their PVCs. We had better wait untilpodCreationTimeout.Does this PR introduce any user-facing change?
This affects only the initial set of executors because the baseline is PVC's
getCreationTimestamp. So, this fixes only a buggy situation where a PVC is shared by two executors due to the long pending executor pod.How was this patch tested?
Pass the CIs with newly updated test cases.
Was this patch authored or co-authored using generative AI tooling?
No.