You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ambari.apache.org by "Hari Sekhon (JIRA)" <ji...@apache.org> on 2018/07/24 09:14:00 UTC
[jira] [Resolved] (AMBARI-24334) HBase Balancer - temporarily
disable balancer for HBase rolling restart
[ https://issues.apache.org/jira/browse/AMBARI-24334?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Hari Sekhon resolved AMBARI-24334.
----------------------------------
Resolution: Not A Problem
Actually don't do this, testing has shown a significant region count and req/sec imbalance from a rolling restart across RegionServers when leaving the balancer disabled during the rolling restart.
It's better to lose some locality than end up with a bigger region and req/sec skew.
The last RegionServer to restart ended up with 0 regions and the first couple ended up with over 250 regions each. Running a quick balancer resolved this, so may as well leave this running through the rolling restart.
> HBase Balancer - temporarily disable balancer for HBase rolling restart
> -----------------------------------------------------------------------
>
> Key: AMBARI-24334
> URL: https://issues.apache.org/jira/browse/AMBARI-24334
> Project: Ambari
> Issue Type: Improvement
> Components: ambari-server, rolling-upgrade
> Affects Versions: 2.5.2
> Reporter: Hari Sekhon
> Priority: Major
>
> Temporarily disable HBase Balancer when doing HBase rolling restarts.
> I've noticed a drop in data locality after doing an HBase rolling restart looking at the Ambari Grafana graph found on the HBase Regionservers dashboard -> Locality (row at bottom) -> Percent Files Local graph.
> I believe the HBase balancer should be temporarily disabled (if enabled) and then re-enabled after all RegionServers have successfully completed. One must be careful to only re-enable the balancer if it was enabled before the rolling restart was initiated.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)