[
https://issues.apache.org/jira/browse/HBASE-9842?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13807665#comment-13807665
]
rajeshbabu commented on HBASE-9842:
-----------------------------------
Thanks Lars for fixing the test case.
The 0.94 patch at HBASE-9593 is hanging because we have only one regionserver
which is getting killed. But during master initialization we will retry
infinitely until root region assigned. Since there are no regionsevers its
hanging there. In the current patch I have added one more region server to open
the -ROOT- and .META. regions. Even though started two regionservers finally
one will be online. without patch two will be online.
bq. Looks like is also wrong on the trunk patch.
In trunk patch starting 1 RS only. After the RS killed there wont be any
regionserver still master initialization not waiting for meta region to assign.
I think the reason for that is HBASE-9151(When meta server znode deleted and
meta in FAILED_OPEN state, then hbck cannot fix it. This scenario can come when
all region servers stopped by stop command and didnt start any RS within 10
secs(with default configurations).)
> Backport HBASE-9593 and HBASE-8667 to 0.94
> ------------------------------------------
>
> Key: HBASE-9842
> URL: https://issues.apache.org/jira/browse/HBASE-9842
> Project: HBase
> Issue Type: Bug
> Reporter: Lars Hofhansl
> Assignee: rajeshbabu
> Fix For: 0.94.14
>
> Attachments: 9842-fixed.txt, HBASE-9842.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.1#6144)
