[
https://issues.apache.org/jira/browse/HBASE-21330?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16654582#comment-16654582
]
Duo Zhang commented on HBASE-21330:
-----------------------------------
No, I've seen it failed several times in the past. Tried locally it passed.
There should be other problems which cause it to fail. Let me commit.
> ReopenTableRegionsProcedure will enter an infinite loop if we schedule a TRSP
> at the same time
> ----------------------------------------------------------------------------------------------
>
> Key: HBASE-21330
> URL: https://issues.apache.org/jira/browse/HBASE-21330
> Project: HBase
> Issue Type: Sub-task
> Components: amv2
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Priority: Major
> Fix For: 3.0.0, 2.2.0
>
> Attachments: HBASE-21330.patch
>
>
> The problem is that, if there are regions which already have TRSPs with them,
> we will give up and just to REOPEN_TABLE_REGIONS_CONFIRM_REOPENED directly.
> And in REOPEN_TABLE_REGIONS_CONFIRM_REOPENED, if there are still regions need
> to be reopened, we will just back to REOPEN_TABLE_REGIONS_REOPEN_REGIONS
> directly. And since ReopenTableRegionsProcedure holds the exclusive lock on
> table, the TRSP will not have chance to execute, and the result is that, the
> ReopenTableRegionsProcedure just keep looping in the two states.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
