[
https://issues.apache.org/jira/browse/HBASE-22709?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16891536#comment-16891536
]
Guanghao Zhang commented on HBASE-22709:
----------------------------------------
bq. But I think this maybe a bit dangerous? If the balancer is still running,
we may introduce more inconsistencies if we do this automatically...
Yes, the check result is not absolutely right. The "fix" thing shoule be better
done by user manually.
And for the inconsistent regions, current HBCK2 is enough to fix them.
Case1. Master thought this region opened, but no regionserver reported it.
====> Fix: Use assign cmd.
Case2. Master thought this region opened on Server1, but regionserver reported
Server2.
====> Fix: need to check the server is still exist. If not, schedule SCP for
it. If exist, restart Server2 and Server1.
Case3. More than one regionservers reported opened this region
====> Fix: restart the regionservers.
> Add a chore thread in master to do hbck checking
> ------------------------------------------------
>
> Key: HBASE-22709
> URL: https://issues.apache.org/jira/browse/HBASE-22709
> Project: HBase
> Issue Type: Sub-task
> Reporter: Guanghao Zhang
> Assignee: Guanghao Zhang
> Priority: Major
> Attachments: HBASE-22709.master.001.patch
>
>
> See HBASE-21965. There may have two type of failed splited/merged regions.
> One is orphan region on filesystem. Another one is unassigned daughter
> regions or merged regions. Plan to add a seperate hbck.jsp for them. Will add
> more hbck problems to this UI in future.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
