[
https://issues.apache.org/jira/browse/HIVE-12581?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15038491#comment-15038491
]
Sergey Shelukhin commented on HIVE-12581:
-----------------------------------------
We are upgrading to DN 4 somewhere, so maybe we can also switch to a new fancy
connection pool after that. Syncing around the connection pool is not really an
option, so I don't know what a workaround would be... connection pool
threadpool where we kill stuck threads?
> Concurrent calls to BoneCP.getConnection() lock up
> --------------------------------------------------
>
> Key: HIVE-12581
> URL: https://issues.apache.org/jira/browse/HIVE-12581
> Project: Hive
> Issue Type: Bug
> Components: Transactions
> Affects Versions: 1.0.0
> Reporter: Eugene Koifman
> Assignee: Eugene Koifman
>
> {noformat}
> Thread 31078: (state = BLOCKED)
> - sun.misc.Unsafe.park(boolean, long) @bci=0 (Interpreted frame)
> - java.util.concurrent.locks.LockSupport.parkNanos(java.lang.Object, long)
> @bci=20, line=226 (Interpreted frame)
> -
> java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(long)
> @bci=68, line=2082 (Interpreted frame)
> - java.util.concurrent.LinkedBlockingQueue.poll(long,
> java.util.concurrent.TimeUnit) @bci=62, line=467 (Interpreted frame)
> - com.jolbox.bonecp.DefaultConnectionStrategy.getConnectionInternal()
> @bci=50, line=82 (Interpreted frame)
> - com.jolbox.bonecp.AbstractConnectionStrategy.getConnection() @bci=6,
> line=90 (Interpreted frame)
> - com.jolbox.bonecp.BoneCP.getConnection() @bci=4, line=553 (Interpreted
> frame)
> - com.jolbox.bonecp.BoneCPDataSource.getConnection() @bci=108, line=131
> (Interpreted frame)
> - org.apache.hadoop.hive.metastore.txn.TxnHandler.getDbConn(int) @bci=16,
> line=973 (Interpreted frame)
> -
> org.apache.hadoop.hive.metastore.txn.TxnHandler.checkLock(org.apache.hadoop.hive.metastore.api.CheckLockRequest)
> @bci=88, line=530 (Interpreted frame)
> {noformat}
> Sometimes these lock up, i.e. some sort of deadlock happens inside the
> connection pool
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
