[
https://issues.apache.org/jira/browse/BOOKKEEPER-500?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13529650#comment-13529650
]
Sijie Guo edited comment on BOOKKEEPER-500 at 12/12/12 6:09 AM:
----------------------------------------------------------------
the users of BOOKKEEPER heavily reply on the correctness of fencing, like
Hedwig replies on BOOKKEEPER fencing to provide right sequence id, NameNode of
HDFS would also need it too. so if the bug affected or would affect the
correctness of fencing, it should be resolved in latest release. so I think it
is a blocker.
was (Author: hustlmsp):
the users of BOOKKEEPER heavily reply on the correctness of fencing, like
Hedwig replies on BOOKKEEPER fencing to provide right sequence id, NameNode of
HDFS would also need it too. so if the bug affects the correctness of fencing,
it should be resolved in latest release. so I think it is a blocker.
> Fencing doesn't work when restarting bookies.
> ---------------------------------------------
>
> Key: BOOKKEEPER-500
> URL: https://issues.apache.org/jira/browse/BOOKKEEPER-500
> Project: Bookkeeper
> Issue Type: Bug
> Components: bookkeeper-server
> Affects Versions: 4.1.0, 4.2.0
> Reporter: Sijie Guo
> Assignee: Sijie Guo
> Priority: Blocker
> Fix For: 4.2.0
>
> Attachments: BOOKKEEPER-500.-testcase.diff
>
>
> Currently we just record fence state in memory. If the bookies are restarted,
> all fence states are gone. It would break the correctness that we guarantee
> about fencing.
> We should record the fence state in the file info to make it persistent.
> A test case added to produce this issue.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
