jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Antonio Martinez-Navarro (JIRA)" <j...@apache.org>
Subject [jira] Commented: (JCR-1801) Support Online Backup
Date Mon, 13 Oct 2008 18:00:44 GMT

    [ https://issues.apache.org/jira/browse/JCR-1801?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12639138#action_12639138

Antonio Martinez-Navarro commented on JCR-1801:

This new feature would be usefull in two scenarios: "Online Backup" and "Online Addition of
Cluster Node".

In both cases the problem is that we need to make a copy of the local data (i.e, repository
dir, revision.log file and workspaces dir), guaranteeing that it is in sync with the DB at
the time of the snapshot. 
The reason why this is needed is is because in large DBs (for example I have tested with 
20G DB), indexing (with local indexing data missing) would take more than 3 hours, and we
can run out of Heap (tested with 2G Heap)

For Online Backup, the user application will backup both the DB and the local index data

For "Online Addition of Cluster Node", the user application will backup ONLY the local data.
This snapshot of local data can then be copied by the user to the new cluster node. When the
new cluster node comes up, it will have to synchronize to the latest snapshot (which can have
changes as the cluster is not stopped), but if the difference is small there should be no

> Support Online Backup
> ---------------------
>                 Key: JCR-1801
>                 URL: https://issues.apache.org/jira/browse/JCR-1801
>             Project: Jackrabbit
>          Issue Type: New Feature
>          Components: clustering, jackrabbit-core
>            Reporter: Thomas Mueller
>            Priority: Minor
> There is no 'good' way to create an online backup for Jackrabbit. Currently you need
to basically stop the repository before backing up the data, or at least make sure in the
application that nobody writes to the repository.
> One solution is to add a feature to block write access to the repository until the backup
is complete.
> Example configuration: database persistence manager, Lucene index, data store.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message