jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marcel Reutegger <marcel.reuteg...@gmx.net>
Subject Re: Several newbie questions:
Date Tue, 09 May 2006 08:32:53 GMT
if the jvm crashes after changes have been persisted in the PersistenceManager but 
before the redo log is written the index will be out-of-date. this is usually just a 
small time frame, but still should be fixed. There is an existing jira report that 
describes this issue:


Michael Baranov wrote:
> hi!
>> jackrabbit uses a redo log to add support for atomic operations to the 
>> lucene index. if the jvm should crash while nodes are added to or 
>> removed from the index then the changes will be rolled back on the 
>> next startup. likewise if changes exist in the redo log but not yet in 
>> the index files and the jvm crashes then the changes are replayed on 
>> the next startup.
> So this technically means if one uses DB backed (say the preferred 
> embedded Derby) and Lucene index altogether, it is very unlikely that 
> the repository becomes damaged, on JVM kill, right? But all through the 
> site docs there are notes that the repository actually may get 
> damaged... So what's in reality? Can you, please, elaborate it a bit here?
> Thanks,
>    Michael.

View raw message