Clustered scheduler service on SwitchYard

[synclpz]

Could anyone propose the way the scheduled service execution could be implemented in SY running in a cluster?

Existing quartz service bindning allows execution of service via specified scheduling expression, but if I run cluster of several SY servers I got this service executed on each node at a given time. I have a need for "clustered" quartz so the service is executed only on one cluster node (e.g. with round-robin behaviour) and the outages of SY node should be detected automatically to allow service to be run on different node if, say, "master" fails...

Any suggestions to implement this via existing tools? Or I should just write my own clustered implementation for quartz?

[kcbabo]

Clustering in quartz can be handled by having individual schedule instances point to the same job store, which essentially synchronizes their activity based on database state.  Here's a bit of background on that:

http://quartz-scheduler.org/documentation/quartz-2.x/configuration/ConfigJDBCJobStoreClustering

The timer configuration for your binding can be set using a quartz.properties file included in the root of your deployment.  The timer name for the quartz binding needs to match the schedule name in your quartz configuration.

If the above doesn't meet your requirements then you are likely looking at setting up your own singleton instance for the cluster to coordinate the timer.

[synclpz]

Nice, but I have only NoSQL DB in my setup

May be I should find some lightweight distributed DB for storing such info, do you know one? H2 is a candidate, but if I only setup it in standalone mode + clustered.