prashantwason commented on code in PR #3207:
URL: https://github.com/apache/hudi/pull/3207#discussion_r850259966
##########
hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/client/SparkRDDWriteClient.java:
##########
@@ -482,4 +483,10 @@ protected void initWrapperFSMetrics() {
HoodieWrapperFileSystem.setMetricsRegistry(registry, registryMeta);
}
}
+
+ @Override
+ protected void releaseResources() {
+ ((HoodieSparkEngineContext)
context).getJavaSparkContext().getPersistentRDDs().values()
Review Comment:
@vinothchandar @minihippo This fix assumes that there is a single
SparkHoodieWriteClient on the JVM. We have a usecase where a single process
handles ingestion for multiple HUDI datasets each having their own
SparkRDDWriteClient in different threads. The first write client which
completes unpersists the RDDs for all other clients causing perf issues.
It may be better to have a write config (default false) for this fix.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
