You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Andrey Stepachev (JIRA)" <ji...@apache.org> on 2014/02/13 17:23:19 UTC

[jira] [Updated] (HIVE-6427) Hive Server2 should reopen Metastore client in case of any Thrift exceptions

     [ https://issues.apache.org/jira/browse/HIVE-6427?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Andrey Stepachev updated HIVE-6427:
-----------------------------------

    Description: 
In case of metastore restart hive server doesn't reopen connection to metastore. Any command gives broken pipe or similar exceptions.
http://paste.ubuntu.com/6926215/

Any other subsequent request doesn't reestablish connection and tries to use stale (closed) connection.

Looks like we shouldn't blindly convert any MetaException to HiveSQLException, but should distinguish fatal exceptoins and logical exceptions.

  was:
In case of metastore restart hive server doesn't reopen connection to metastore. Any command gives broken pipe or similar exceptions.
http://paste.ubuntu.com/6926215/

Any other subsequent requests doesn't reestablish connection and tries to use stale (closed) connection.

Looks like we shouldn't blindly convert any MetaException to HiveSQLException, but should distinguish fatal exceptoins and logical exceptions.


> Hive Server2 should reopen Metastore client in case of any Thrift exceptions
> ----------------------------------------------------------------------------
>
>                 Key: HIVE-6427
>                 URL: https://issues.apache.org/jira/browse/HIVE-6427
>             Project: Hive
>          Issue Type: Bug
>          Components: HiveServer2
>    Affects Versions: 0.12.0
>         Environment: cloudera cdh5 beta2
>            Reporter: Andrey Stepachev
>            Priority: Critical
>
> In case of metastore restart hive server doesn't reopen connection to metastore. Any command gives broken pipe or similar exceptions.
> http://paste.ubuntu.com/6926215/
> Any other subsequent request doesn't reestablish connection and tries to use stale (closed) connection.
> Looks like we shouldn't blindly convert any MetaException to HiveSQLException, but should distinguish fatal exceptoins and logical exceptions.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)