Hi all,
I found a issue that I probably have to solve for a project:
http://jira.magnolia-cms.com/browse/MGNLSCH-15
I'm just looking for idea on how to solve it. I mean, the part about
executing a job on certain nodes seems pretty easy, you can set a
property with the node names and then compare it with magnolia.clusterid
(or the jackrabbit one) property that is already guaranteed to be unique
in the cluster.
But I can't see an easy solution for the part about running in just one
node. Ok, you can set a single cluster id in the aforementioned idea,
but this means that if the node you selected is down, the job is lost,
which is pretty much what I want to avoid with the cluster.
Also there is no "schedule" workspace that could be used, so I thought
to use a lock on some node (maybe the node defining the job itself in
config), but I have no idea if it's possible to set a cluster-wise lock
on a node, how to check if it's currently locked and how eventually
force the unlock if the cluster node has died for some reason. Have you
any pointer (with code samples) on these topics? Thank you.
Regards, Danilo.
----------------------------------------------------------------
For list details see
http://www.magnolia-cms.com/community/mailing-lists.html
To unsubscribe, E-mail to: <[email protected]>
----------------------------------------------------------------
