Skip to content

[SPARK-37688][CORE] ExecutorMonitor should ignore SparkListenerBlockUpdated event if executor was not active#34956

Closed
sleep1661 wants to merge 2 commits intoapache:masterfrom
sleep1661:SPARK-37688
Closed

[SPARK-37688][CORE] ExecutorMonitor should ignore SparkListenerBlockUpdated event if executor was not active#34956
sleep1661 wants to merge 2 commits intoapache:masterfrom
sleep1661:SPARK-37688

Conversation

@sleep1661
Copy link
Contributor

What changes were proposed in this pull request?

ExecutorMonitor should ignore SparkListenerBlockUpdated event if executor was not active

Why are the changes needed?

If not ignored, ExecutorMonitor will create a new executor tracker with UNKNOWN_RESOURCE_PROFILE_ID for the dead executor. And ExecutorAllocationManager will not remove executor with UNKNOWN_RESOURCE_PROFILE_ID, which cause a executor slot is occupied by the dead executor, so a new one cannot be created.

Does this PR introduce any user-facing change?

No

How was this patch tested?

Add a new test.

@github-actions github-actions bot added the CORE label Dec 20, 2021
@AmplabJenkins
Copy link

Can one of the admins verify this patch?

@sleep1661
Copy link
Contributor Author

cc @mridulm

@mridulm
Copy link
Contributor

mridulm commented Dec 28, 2021

Looks fine to me.
+CC @Ngone51 since you reviewed the original PR and have more context here - any concerns with the increased lock contention in CoarseGrainedSchedulerBackend ?

@sleep1661
Copy link
Contributor Author

ping @Ngone51

@sleep1661
Copy link
Contributor Author

@cloud-fan Could you help take a look when you have time? Thanks.

@asfgit asfgit closed this in 969d672 Mar 1, 2022
@mridulm
Copy link
Contributor

mridulm commented Mar 1, 2022

Merged to master.
Thanks for working on this @sleep1661 !

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants

Comments