[
https://issues.apache.org/jira/browse/HIVE-7021?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Naveen Gangam updated HIVE-7021:
--------------------------------
Attachment: HIVE-4629+HIVE-7021.1.patch
I am also attaching the full patch that contains the fixes from both HIVE-4629
+ HIVE-7021 build from the trunk today.
> HiveServer2 memory leak on failed queries
> -----------------------------------------
>
> Key: HIVE-7021
> URL: https://issues.apache.org/jira/browse/HIVE-7021
> Project: Hive
> Issue Type: Bug
> Components: HiveServer2
> Affects Versions: 0.12.0
> Reporter: Naveen Gangam
> Assignee: Naveen Gangam
> Fix For: 0.14.0
>
> Attachments: HIVE-4629+HIVE-7021.1.patch, HIVE-7021.1.patch
>
>
> The number of the following objects keeps increasing if a query causes an
> exception:
> org.apache.hive.service.cli.HandleIdentifier
> org.apache.hive.service.cli.OperationHandle
> org.apache.hive.service.cli.log.LinkedStringBuffer
> org.apache.hive.service.cli.log.OperationLog
> The leak can be observed using a JDBCClient that runs something like this
> connection =
> DriverManager.getConnection("jdbc:hive2://" + hostname + ":10000/default",
> "", "");
> statement = connection.createStatement();
> statement.execute("CREATE TEMPORARY FUNCTION
> dummy_function AS 'dummy.class.name'");
> The above SQL will fail if HS2 cannot load "dummy.class.name" class. Each
> iteration of such query will result in +1 increase in instance count for the
> classes mentioned above.
> This will eventually cause OOM in the HS2 service.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
