jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dominique Pfister" <dominique.pfis...@day.com>
Subject Re: ClusterNode and snapshot question.
Date Thu, 22 May 2008 06:25:38 GMT
Hi Ian,

To bring up a new node into a cluster, I'd do the following:

(1) Stop an already existing node
(2) Make a deep copy of the existing node, starting from the repository home
(3) Assign a unique cluster node id to this new node
(4) Startup both nodes again

Stopping the source node before copying ensures that indices are
consistent. In (2) the revision.log will get copied as well, so the
new cluster node will start from the same revision as the source node.

Kind regards

On 5/21/08, Ian Boston <ieb@tfd.co.uk> wrote:
> I have a question about ClusterNode and the Journal.
>  If I bring a new app server node into a Jackrabbit cluster, the journal
> gets replayed from scratch, this can take some time if the cluster has been
> running for some time. I would like to avoid this if possible.
>  It would be nice to be able to snapshot the local state of an app server,
> and use that to a) seed new app servers as they joined the cluster b)
> truncate the redo log to prevent eternal growth.
>  Is this the right way to achieve a fast start of a new app server in the
> cluster ?
>  Is it necessary to run the journal from scratch (I assume it is due to the
> indexing operations)?
>  Has anyone already looked at this ?
>  I searched the lists but I couldn't find anything.
>  Thanks
>  Ian

View raw message