[jira] Updated: (HBASE-3650) HBA.delete can return too fast

["Jean-Daniel Cryans (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HBASE-3650?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jean-Daniel Cryans updated HBASE-3650:
--------------------------------------

    Attachment: HBASE-3650.patch

This patch adds a new utility in MetaReader to configure a Scan that only reads the rows from a single table, then refactors both HBA and MetaReader to use it. Finally it adds a test to make sure the method is getting the right rows from .META.

> HBA.delete can return too fast
> ------------------------------
>
>                 Key: HBASE-3650
>                 URL: https://issues.apache.org/jira/browse/HBASE-3650
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 0.90.1
>            Reporter: Jean-Daniel Cryans
>            Assignee: Jean-Daniel Cryans
>            Priority: Blocker
>             Fix For: 0.90.2
>
>         Attachments: HBASE-3650.patch
>
>
> One of our engineers got a weird TableExistsException in his code and I see that the client-side logging says the table was deleted in less than a second while it took the master 5 seconds to do it. Doing code inspection, the .META. scanner in HBA.delete can set _found_  to _true_ and then set it back to _false_ in the case where the deleted table isn't the last one. We should just do a scan that would get the rows specific to the table instead of scanning *all* the rows.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira