[
https://issues.apache.org/jira/browse/HBASE-13895?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14589083#comment-14589083
]
stack commented on HBASE-13895:
-------------------------------
Pardon me [~enis] I missed this. Indeed I had v1 of HBASE-13877 in place.
I look to have something else though. Master crashes and the new, active master
assigns a region already happy out on the cluster. Feel free to take this issue
Enis since you dumped in your sequel already. I'll go write my novel in a new
JIRA.
Here is the trailer:
Region 6fbe22ff15c2e5f2b207f79eaf8f382a was successfully opened on c2025.
{code}
2015-06-09 20:05:37,661 DEBUG [AM.ZK.Worker-pool2-t27] master.RegionStates:
Onlined 6fbe22ff15c2e5f2b207f79eaf8f382a on
c2025.halxg.cloudera.com,16020,1433892619022
...
2015-06-09 20:05:37,962 DEBUG
[c2020.halxg.cloudera.com,16000,1433905379975-GeneralBulkAssigner-2]
master.AssignmentManager: Bulk assigning done for
c2025.halxg.cloudera.com,16020,1433892619022
{code}
Master crashes and is restarted a few seconds later... in comes c2025 and
registers
{code}
2015-06-09 20:06:01,864 INFO [main] util.VersionInfo: HBase 1.2.0-SNAPSHOT
....
2015-06-09 20:06:16,677 INFO [PriorityRpcServer.handler=10,queue=0,port=16000]
master.ServerManager: Registering
server=c2025.halxg.cloudera.com,16020,1433892619022
...
2015-06-09 20:06:20,026 INFO [c2020:16000.activeMasterManager]
master.MasterFileSystem: Log folder
hdfs://c2020.halxg.cloudera.com:8020/hbase/WALs/c2025.halxg.cloudera.com,16020,1433892619022
belongs to an existing region server
{code}
I think it has something to do w/ the fact that c2025 is carrying meta....
still digging.
> DATALOSS: Double-assignment ITBLL'ing
> -------------------------------------
>
> Key: HBASE-13895
> URL: https://issues.apache.org/jira/browse/HBASE-13895
> Project: HBase
> Issue Type: Bug
> Affects Versions: 1.2.0
> Reporter: stack
> Assignee: stack
> Priority: Critical
> Fix For: 1.2.0, 1.1.1
>
> Attachments: hbase-13895_v1-branch-1.1.patch
>
>
> Opening a place holder till finish analysis.
> I have dataloss running ITBLL at 3B (testing HBASE-13877). Most obvious
> culprit is the double-assignment that I can see.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
