[
https://issues.apache.org/jira/browse/HBASE-7966?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13592646#comment-13592646
]
Enis Soztutar commented on HBASE-7966:
--------------------------------------
bq. Yes i agree with this. I did not mean to revert the table lock fix. I just
meant to say with this fix we have exposed some internal problem, may be the
wait logic can be improved in the lines of the new fix.
I was also suspecting that acquiring the table lock might be stalling table
creation a bit more, since we have to do create znode + list children for
acquiring the table lock even without contention. But could not look into that
last week. Though it should not take multiple seconds.
Thanks for looking into this.
> ACL tests fail on trunk (flaky)
> -------------------------------
>
> Key: HBASE-7966
> URL: https://issues.apache.org/jira/browse/HBASE-7966
> Project: HBase
> Issue Type: Bug
> Components: security
> Reporter: Enis Soztutar
> Assignee: stack
> Fix For: 0.95.0, 0.98.0
>
> Attachments: 7966-v1.txt, 7966v3.txt, 7966v4.txt, debug.logging.txt
>
>
> I've noticed that these tests fail for me quite frequently:
> {code}
> org.apache.hadoop.hbase.security.access.TestTablePermissions:
> org.apache.hadoop.hbase.exceptions.TableNotFoundException: _acl_
>
> testQualifierAccess(org.apache.hadoop.hbase.security.access.TestAccessControlFilter):
> org.apache.hadoop.hbase.exceptions.TableNotFoundException: _acl_
> {code}
> The root cause seems to be that, the AccessController only creates the table
> in HMaster.postStartMaster(), but before _acl_ is created, master start
> accepting other create table requests, hence race.
--
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
