[
https://issues.apache.org/jira/browse/HBASE-4182?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13081556#comment-13081556
]
ramkrishna.s.vasudevan commented on HBASE-4182:
-----------------------------------------------
{noformat}
2011-08-09 16:04:57,429 INFO org.apache.hadoop.hbase.master.LoadBalancer:
Calculated a load balance in 1ms. Moving 3 regions off of 1 overloaded servers
onto 1 less loaded servers
2011-08-09 16:04:57,429 INFO org.apache.hadoop.hbase.master.HMaster: balance
hri=t5,,1312885663262.e59e66fa2535da2a2feccdffbf22063c.,
src=linux76,60020,1312885513666, dest=linux146,60020,1312885914318
2011-08-09 16:05:06,832 DEBUG org.apache.hadoop.hbase.master.AssignmentManager:
Starting unassignment of region
t5,,1312885663262.e59e66fa2535da2a2feccdffbf22063c. (offlining)
2011-08-09 16:05:49,343 ERROR org.apache.hadoop.hbase.master.HMaster$1: Caught
exception
java.lang.NullPointerException
at
org.apache.hadoop.hbase.master.AssignmentManager.unassign(AssignmentManager.java:1563)
at
org.apache.hadoop.hbase.master.AssignmentManager.unassign(AssignmentManager.java:1527)
at
org.apache.hadoop.hbase.master.AssignmentManager.balance(AssignmentManager.java:2535)
at org.apache.hadoop.hbase.master.HMaster.balance(HMaster.java:832)
at org.apache.hadoop.hbase.master.HMaster$1.chore(HMaster.java:704)
at org.apache.hadoop.hbase.Chore.run(Chore.java:66)
{noformat}
> NullPointerException when loadbalancer tries to close the region for
> reassigning to new RS.
> -------------------------------------------------------------------------------------------
>
> Key: HBASE-4182
> URL: https://issues.apache.org/jira/browse/HBASE-4182
> Project: HBase
> Issue Type: Bug
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
>
> 1. Start 2 RS. Create some regions so that is is balanced.
> 2. Stop RS2. Now all the Regions from RS2 are assigned to RS1.
> 3. Again start RS2.
> 4. Load Balancing is calculated and few regions from RS1 are assigned to RS2.
> As part of this step Master tries to unassign the regions from RS1.
> {noformat}
> RegionTransitionData data = ZKAssign.getDataNoWatch(zkw, ZKAssign
> .getNodeName(zkw, region.getEncodedName()), null);
> if (data.equals(EventType.RS_ZK_REGION_CLOSING)) {
> ZKAssign.createNodeClosing(zkw, region, master.getServerName());
> }
> {noformat}
> Now there is no data present in the unassigned node. We are directly
> comparing the data.
> Here data is null. Hence nullpointer exception is thrown.
> Hence load balancing fails.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
