[
https://issues.apache.org/jira/browse/HBASE-11405?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sean Busbey updated HBASE-11405:
--------------------------------
Attachment: HBASE-11405-0.94.patch.txt
HBASE-11405-0.98.patch.txt
Here's an amended version specific to each of 0.98 and 0.94 that works under
profiles for hadoop 1.0, 1.1, and 2.x. Due to the amount of change, I added an
Amending-Author to the commit message.
I've run through {{-Dtest=TestHBaseFsck}} for the above profiles and
{{-PrunSmallTests}} for whatever the default profile is on each branch. NB
running TestHBaseFsck on 0.94 required {{-PlocalTests}}.
Fixing this involved exposing a previously private method for
HBaseTestingUtility under 0.94. Since nothing else needed it in 0.98 I marked
it deprecated with a note that it would disappear with Hadoop 1.x support.
Since the method isn't in branch-1 or later, that should work out well.
One worry: some of the TestHBaseFsck tests get a bit flaky on 0.94 with this
change on my laptop (which admittedly is underpowered). However, even leaving
in a single retry period causes the parallel test to fail because the first
invocation finished in < 1 sec.
> Multiple invocations of hbck in parallel disables balancer permanently
> -----------------------------------------------------------------------
>
> Key: HBASE-11405
> URL: https://issues.apache.org/jira/browse/HBASE-11405
> Project: HBase
> Issue Type: Bug
> Components: Balancer, hbck
> Affects Versions: 0.99.0
> Reporter: bharath v
> Assignee: bharath v
> Fix For: 2.0.0, 0.98.7, 0.99.1, 0.94.25
>
> Attachments: 11405-1.0.txt, 11405-v3.txt, 11405-v5.patch,
> 11405-v6.patch, 11405v7.patch, HBASE-11405-0.94.patch.txt,
> HBASE-11405-0.98.patch.txt, HBASE-11405-trunk-rebased.patch,
> HBASE-11405-trunk.patch, HBASE-11405-trunk.patch.1,
> hbase-11405.rebase.140911.patch
>
>
> This is because of the following piece of code in hbck
> {code:borderStyle=solid}
> boolean oldBalancer = admin.setBalancerRunning(false, true);
> try {
> onlineConsistencyRepair();
> }
> finally {
> admin.setBalancerRunning(oldBalancer, false);
> }
> {code}
> Newer invocations set oldBalancer to false as it was disabled by previous
> invocations and this disables balancer permanently unless its manually turned
> on by the user. Easy to reproduce, just run hbck 100 times in a loop in 2
> different sessions and you can see that balancer is set to false in the
> HMaster logs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
