[
https://issues.apache.org/jira/browse/HBASE-21217?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16627167#comment-16627167
]
Allan Yang commented on HBASE-21217:
------------------------------------
{quote}
There is no FAILED_CLOSE state so we can not tell master that the closing is
failed. And I think the only possible way to meet the null region is that, we
have already sent the request to RS, and RS has finished the closing, but a
retrying rpc call has aleady been sent, and finally it is scheduled after we
finish everything, then here we will meet a null region. For this case I think
it is fine to just ignore it? Not sure if there are other possible ways to
enter here, if so I think there will be bugs...
{quote}
Quote from the reviewborad.
Yes, I have encountered this in ITBLL, that's why I want to change back to
CompatRemoteProcedureResolver in branch-2.0 and branch-2.1. But it is
definitely another bug need to reveal and fix.
> Revisit the executeProcedure method for open/close region
> ---------------------------------------------------------
>
> Key: HBASE-21217
> URL: https://issues.apache.org/jira/browse/HBASE-21217
> Project: HBase
> Issue Type: Sub-task
> Components: amv2, proc-v2
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Priority: Critical
> Fix For: 3.0.0, 2.2.0
>
> Attachments: HBASE-21217-v1.patch, HBASE-21217-v2.patch,
> HBASE-21217.patch
>
>
> Currently we just call openRegion and closeRegion directly, which is a bit
> buggy. For example, in order to not fail all the open region requests while
> there is only one failure, we will catch the exception and set a flag in the
> return value. But for executeProcedures call, the return value will be
> ignored, and we expect the openRegion method will always call
> reportRegionStateTransition to report the failure but in fact it does not...
> And after HBASE-20881, we can confirm that the race could happen, where we
> send a close request to a region which is opening(HBASE-21199), and vice
> visa. So I think here we need to revisit the implementation of
> executeProcedures to make it more stable.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
