[
https://issues.apache.org/jira/browse/DIRSERVER-1790?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13554912#comment-13554912
]
Emmanuel Lecharny commented on DIRSERVER-1790:
----------------------------------------------
Ok, I think the pb occurs when the server is shutdown 'brutally'.
We have a shutdown hook in the server, which flush all the partitions on disk
when the server is shutdown using a kill -TERM or a ^C. It's called, and in
this case, all the *.db files are correctly created and filled. Otherwise,
doing a kill -9 will let the index in a intermediate state, where the *.lg
files contain all the updates, but the *.db files all have a 0 size.
A workaround for the moment is to be sure to correctly stop the server using a
kill -TERM on linux, or when stopping the server.
Still, we should periodically flush the journal files (*.lg) so that the data
files (*.db) are updated, and more important, detect that we are in a transient
state when the server is started, and update the data files, applying the
journal on them. This would solve the pb we will face in case of a server
crash. This is definitively needed.
> When we restart the server, some user index are lost
> ----------------------------------------------------
>
> Key: DIRSERVER-1790
> URL: https://issues.apache.org/jira/browse/DIRSERVER-1790
> Project: Directory ApacheDS
> Issue Type: Bug
> Affects Versions: 2.0.0-M9
> Reporter: Emmanuel Lecharny
> Assignee: Emmanuel Lecharny
> Priority: Blocker
> Fix For: 2.0.0-M10
>
>
> Indexes are not anymore used when the server is restarted. The index was
> available when the server was started the first time, then after a restart,
> even if the index exists on disk, it's not anymore used by the server,
> causing a full scan to be done.
--
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
