hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thejas M Nair (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-9642) Hive metastore client retries don't happen consistently for all api calls
Date Thu, 26 Feb 2015 21:43:04 GMT

    [ https://issues.apache.org/jira/browse/HIVE-9642?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14339215#comment-14339215
] 

Thejas M Nair commented on HIVE-9642:
-------------------------------------

We should also add a unit test for this. ie, -
 # bring up two metstore servers
 # make an api call
 # bring down first server (thats one it connects to)
 # make the api call again
 # bring down 2nd server 
 # bring up first server
 # make the api call again

See TestHiveMetaStorePartitionSpecs.RunMS on starting metastore server. 


> Hive metastore client retries don't happen consistently for all api calls
> -------------------------------------------------------------------------
>
>                 Key: HIVE-9642
>                 URL: https://issues.apache.org/jira/browse/HIVE-9642
>             Project: Hive
>          Issue Type: Bug
>    Affects Versions: 1.0.0
>            Reporter: Xiaobing Zhou
>            Assignee: Xiaobing Zhou
>         Attachments: HIVE-9642.1.patch, HIVE-9642.2.patch
>
>
> When org.apache.thrift.transport.TTransportException is thrown for issues like socket
timeout, the retry via RetryingMetaStoreClient happens only in certain cases.
> Retry happens for the getDatabase call in but not for getAllDatabases().
> The reason is RetryingMetaStoreClient checks for TTransportException being the cause
for InvocationTargetException. But in case of some calls such as getAllDatabases in HiveMetastoreClient,
all exceptions get wrapped in a MetaException. We should remove this unnecessary wrapping
of exceptions for certain functions in HMC.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message