[
https://issues.apache.org/jira/browse/HBASE-1082?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-1082:
-------------------------
Attachment: 1082.patch
First crack at this functionality. Adds flush and major compaction to shell.
Also added compact and split to the hbase shell while I was at it:
{code}
hbase(main):003:0> tools
HBASE SURGERY TOOLS:
close_region Close a single region. Optionally specify regionserver.
Examples:
hbase> close_region 'REGIONNAME'
hbase> close_region 'REGIONNAME', 'REGIONSERVER_IP:PORT'
disable_region Disable a single region
enable_region Enable a single region. For example,
hbase> enable_region 'REGIONNAME'
flush Flush all regions in passed table or pass a region row to
flush an individual region
hbase> flush 'TABLENAME'
hbase> flush 'REGIONNAME'
compact Compact all regions in passed table or pass a region row
to compact an individual region
major_compact Run major compaction on passed table or pass a region row
to major compact an individual region
split Split table or pass a region row to split individual region
Above commands are for 'experts'-only as misuse can damage an install
{code}
> Administrative functions for table/region maintenance
> -----------------------------------------------------
>
> Key: HBASE-1082
> URL: https://issues.apache.org/jira/browse/HBASE-1082
> Project: Hadoop HBase
> Issue Type: New Feature
> Affects Versions: 0.19.0
> Reporter: Jonathan Gray
> Fix For: 0.19.0
>
> Attachments: 1082.patch
>
>
> It would be useful to have some administrative functions available through
> HTable or HBaseAdmin. The two functions I'm thinking of right now are to
> force memcache flushes (on all regions, or individually specified) and to run
> major compactions (on all regions, or individually specified).
> One reason to have this is currently major compactions run once a day by
> default. The time that they run is related to when you brought up your
> cluster or created the tables. In my case, this has been during peak load
> rather than offpeak times. We have times that we run other administrative
> tasks like backups, cleanup, etc... during off times, and this would be a
> good time to also trigger major compactions and memcache flushes.
> Memcache force flushing can also be useful in the case that your cluster
> starts to have issues. They might be isolated to a single region or
> regionserver, but with lots of edits sitting in memcache and potentially
> unappended hlogs, you want to just flush things out to remove the possibility
> of losing anything.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
