Drools failover/clustering strategies

[gmh]

Hi all,

I am curious as to how one can set up a failover/clustering strategy as a part of disaster recovery with Drools.

Consider the following common scenerio:

1) One Drools server goes down and the state need to be persisted and the processing need to be interruption-proof.

     How should one implement a strategy for it?

          Two execution servers with the same the same facts. As soon as one goes down the other one picks it up? How can that be implemented? via workbench or                       programmatically?

     How should one create a cluster of Drools servers? 

          Use Workbench? Can it be done programmatically?

Ideas? suggestions?

Any sample projects/code that can be shared with me?

Thanks, Gordon

[Marian Buenosayres]

I think Duncan Doyle started working on something like that. You can find his presentation about it here: https://rhsummit.files.wordpress.com/2014/04/doyle_h_0945-high-availablity-cep-with-red_hat-jboss_brms1.pdf

Also, I wrote something on the topic which might be a good start if you're familiar with KieSession configs: https://marianbuenosayres.wordpress.com/2013/07/10/making-a-non-persistent-ha-knowledge-session/

Cheers,

Mariano

[Yun Lin]

Hi the jars in the  https://marianbuenosayres.wordpress.com/2013/07/10/making-a-non-persistent-ha-knowledge-session/ are broken. Could you fix it?