Tao Li commented on HIVE-15002:

Thanks for the patch. Quick comments:

1. Moving the operation creating in the try block makes sense to me. But did 
you see exceptions thrown when creating a ExecuteStatementOperation? Just want 
to make sure we did run into this situation in reality even though it's rare.

2. nit: why did you remove "OperationManager operationManager = 
getOperationManager();" and call getOperationManager() instead? I don't think 
that's really necessary.

> HiveSessionImpl#executeStatementInternal may leave locks in an inconsistent 
> state
> ---------------------------------------------------------------------------------
>                 Key: HIVE-15002
>                 URL: https://issues.apache.org/jira/browse/HIVE-15002
>             Project: Hive
>          Issue Type: Bug
>          Components: HiveServer2
>            Reporter: Zoltan Haindrich
>            Assignee: Zoltan Haindrich
>         Attachments: HIVE-15002.1.patch
> {{TestEmbeddedThriftBinaryCLIService}} have showed me this behaviour.
> Thrift service may hang because of this issue.
> {{TestEmbeddedThriftBinaryCLIService}} can reproduce this problem with a very 
> bad probability...but I was able to 
> Its not able to aquire the semaphore; because most probably because an 
> earlier command have produced an SQLException in the unprotected parts.

This message was sent by Atlassian JIRA

Reply via email to