perf: reuse CometConf.COMET_TRACING_ENABLED, Native, NativeUtil in NativeBatchDecoderIterator#3627
Merged
andygrove merged 1 commit intoapache:mainfrom Mar 3, 2026
Conversation
…reShuffleReader that can be reused.
andygrove
approved these changes
Mar 3, 2026
Contributor
kazuyukitanimura
left a comment
There was a problem hiding this comment.
Thanks @mbutrovich
|
|
||
| if (taskContext != null) { | ||
| taskContext.addTaskCompletionListener[Unit](_ => { | ||
| close() |
Contributor
There was a problem hiding this comment.
@mbutrovich is this close get called somewhere else? Just making sure, because the new code does not have this clause anymore.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Which issue does this PR close?
N/A.
Rationale for this change
JFR profiling shows ~1.5% of on-CPU time in
NativeBatchDecoderIterator.<init>, mostly inSQLConf.get()called fromCometConf.COMET_TRACING_ENABLED.get().SQLConf.getis surprisingly expensive: it reads a thread-local, checks whether a conf is set via string key lookups in the underlying HadoopConfiguration(which holds aPropertiesmap), and may trigger synchronized access. This constructor runs once per shuffle block, so for high-partition shuffles with thousands of blocks per task, the cost accumulates.Additionally,
Native(stateless JNI stub) andNativeUtil(allocates aCDataDictionaryProviderbacked by native memory) are created and destroyed per block. These didn’t show up a ton in the profiling, but might as well hoist it out as well.What changes are included in this PR?
Hoist
Native,NativeUtil, andtracingEnabledout ofNativeBatchDecoderIteratorintoCometBlockStoreShuffleReader.read(), so they are created once per task instead of once per shuffle block.NativeUtilis now closed in the reader's task completion listener.How are these changes tested?
Existing tests.