[
https://issues.apache.org/jira/browse/HIVE-26495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17585622#comment-17585622
]
Naresh P R edited comment on HIVE-26495 at 8/26/22 9:52 PM:
------------------------------------------------------------
[~srahman] I dint see much improvement. I was missing HIVE-21040 in my internal
branch which helped to improve performance in S3 table with millions of files
under partition path.
But still i wanted to pursue with the patch. Let me know if you can review it.
was (Author: nareshpr):
[~srahman] I dint see much improvement. I was missing HIVE-21040 which helped
to improve performance in S3 table with millions of files under partition path.
But still i wanted to pursue with the patch. Let me know if you can review it.
> MSCK repair perf issue HMSChecker ThreadPool is blocked at fs.listStatus
> ------------------------------------------------------------------------
>
> Key: HIVE-26495
> URL: https://issues.apache.org/jira/browse/HIVE-26495
> Project: Hive
> Issue Type: Bug
> Reporter: Naresh P R
> Assignee: Naresh P R
> Priority: Major
> Labels: pull-request-available
> Time Spent: 1h 20m
> Remaining Estimate: 0h
>
> With hive.metastore.fshandler.threads = 15, all 15 *MSCK-GetPaths-xx* are
> slogging at following trace.
> {code:java}
> "MSCK-GetPaths-11" #12345 daemon prio=5 os_prio=0 tid= nid= waiting on
> condition [0x00007f9f099a6000]
> java.lang.Thread.State: WAITING (parking)
> at sun.misc.Unsafe.park(Native Method)
> - parking to wait for <0x00000003f92d1668> (a
> java.util.concurrent.CompletableFuture$Signaller)
> at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
> at
> java.util.concurrent.CompletableFuture$Signaller.block(CompletableFuture.java:1707)
> at java.util.concurrent.ForkJoinPool.managedBlock(ForkJoinPool.java:3323)
> ...
> at org.apache.hadoop.fs.s3a.S3AFileSystem.listStatus(S3AFileSystem.java:3230)
> at org.apache.hadoop.fs.FileSystem.listStatus(FileSystem.java:1953)
> at org.apache.hadoop.fs.FileSystem.listStatus(FileSystem.java:1995)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStoreChecker$PathDepthInfoCallable.processPathDepthInfo(HiveMetaStoreChecker.java:550)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStoreChecker$PathDepthInfoCallable.call(HiveMetaStoreChecker.java:543)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStoreChecker$PathDepthInfoCallable.call(HiveMetaStoreChecker.java:525)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:750){code}
> We should take advantage of non-block listStatusIterator instead of
> listStatus which is a blocking call.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
