hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "rohithsharma (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-3469) When new hive cluster is started, hive jdbc client hangs.
Date Tue, 18 Sep 2012 05:09:07 GMT

    [ https://issues.apache.org/jira/browse/HIVE-3469?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13457605#comment-13457605
] 

rohithsharma commented on HIVE-3469:
------------------------------------

bq. Any idea why this doesn't solve the problem?
I am using hive branch-0.9.0. The above problem is fixed in Hive-Trunk only.

I gone through the hive-trunk code for the above problem. I have one doubt on the synchronization,
Exact synchronization problem exist on method createDefaultDB_core(). Why it is synchronized
for calling method createDefaultDB().?

Correct me If I am wrong, synchronizing calling method impact would be on getting client connection
is delayed. There is "if" block in method "createDefaultDB()"  that check for flag "createDefaultDB".
Synchronizing to this method lead execute this if check only one client. This if check we
can do parallel by bringing down synchronization one layer to the method createDefaultDB_core().

                
> When new hive cluster is started, hive jdbc client hangs.
> ---------------------------------------------------------
>
>                 Key: HIVE-3469
>                 URL: https://issues.apache.org/jira/browse/HIVE-3469
>             Project: Hive
>          Issue Type: Bug
>          Components: Clients, Metastore, Thrift API
>         Environment: Hadoop-2.0.1 + Hive-0.9.0
>            Reporter: rohithsharma
>
> The scenario is as follows  : 
> For the new hive cluster , when multiple hive client is getting the connection for hive
server , there are 3 problems in metastore , hive client.
> 1. Since the derby is booted at first connection,  multiple client try to create database.One
of the client succeed and another one throw exception because of *unique primary constraint*
in metatables.
> 2. Once the Runtime exception is throw, the hive client which is made connection request
will hang forever.
> 3. Potentially "ConcurrentModificationException" is thrown by JDOQuery.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message