[
https://issues.apache.org/jira/browse/HBASE-16910?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16534754#comment-16534754
]
Monani Mihir commented on HBASE-16910:
--------------------------------------
Hi [~mantonov] , are you planning to pick this patch for branch-1.3 ?
Apart from issue mention in JIRA, when ever we do master failover, below error
will be thrown :-
{code:java}
2018-03-10 01:08:37,403 DEBUG [main-EventThread] master.ActiveMasterManager -
No master available. Notifying waiting threads
.
.
.
2018-03-10 01:09:42,571 ERROR [.activeMasterManager]
balancer.StochasticLoadBalancer - failed to get the size of all tables,
exception = null
{code}
Master failover will works without this patch too. Its just throw above
mentioned error even if it is set is *setByTable = false*
> Avoid NPE when starting StochasticLoadBalancer
> ----------------------------------------------
>
> Key: HBASE-16910
> URL: https://issues.apache.org/jira/browse/HBASE-16910
> Project: HBase
> Issue Type: Bug
> Components: Balancer
> Affects Versions: 2.0.0
> Reporter: Guanghao Zhang
> Assignee: Guanghao Zhang
> Priority: Minor
> Fix For: 1.4.0, 2.0.0
>
> Attachments: HBASE-16910-branch-1.patch, HBASE-16910-v1.patch,
> HBASE-16910.patch
>
>
> When master start, it initialize StochasticLoadBalancer.
> {code}
> this.balancer.setClusterStatus(getClusterStatus());
> this.balancer.setMasterServices(this);
> {code}
> It first setClusterStatus(), then setMasterService(). But in setClusterStatus
> method, it use master service which is not initialized. So it will throw NPE.
> {code}
> int tablesCount = isByTable ? services.getTableDescriptors().getAll().size()
> : 1;
> {code}
> It happens when set isByTable is ture.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
