[
https://issues.apache.org/jira/browse/HBASE-10401?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13879153#comment-13879153
]
Jonathan Hsieh commented on HBASE-10401:
----------------------------------------
This was tested manually on a test cluster:
* created 500 regions with data
* moved them off to another dir
* drop then create again with 500 regions with different split points.
* copy the first table's into region dirs.
* run hbck. We end up with 100 or so overlap groups.
* run hbase hbck -fixOverlaps.
* eventually we get back to a consistent state.
> [hbck] perform overlap group merges in parallel
> -----------------------------------------------
>
> Key: HBASE-10401
> URL: https://issues.apache.org/jira/browse/HBASE-10401
> Project: HBase
> Issue Type: Bug
> Components: hbck
> Affects Versions: 0.92.2, 0.98.0, 0.94.16, 0.99.0, 0.96.1.1
> Reporter: Jonathan Hsieh
> Assignee: Jonathan Hsieh
> Attachments: hbase-10401.10400.patch, hbase-10401.patch
>
>
> In a recent support case, we encountered a corrupt hbase that had thousands
> of "overlap groups" (regions that had overlapping key ranges). The current
> implementation repairs these by serially taking a group, perorming a merge
> and then moving on to the next group. Because assignments and hdfs nn
> operations are involved each merge could take on the order of seconds. With
> thousands of overlap groups, this could take hours to complete.
> This patch makes it so that these independent merge groups are merged in
> parallel. It uses the same thread pool for other fs info-gathering
> operations.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
