IT Process Automation

Pam 4.3.1.  I'm having an issue where in a clustered configuration scheduled processes execute twice.  The Orchestrators are behind a Brocade load balancer in an Active / Passive configuration.  When the primary goes offline the secondary is promoted to primary until it goes offline.  

I've got a few processes that generate reports and email them off and when the schedule executes it kicks off two processes, one on each of the Orchestrators.  The only way I've found to prevent multiple emails is to remove the Email connector settings on the passive Orchestrator.  The process still executes twice but one of them fails so the email doesn't get sent.  However if the server with the email connector settings blanked out becomes the primary then both instances of the process fail.  It's pretty weird behavior and I'm not sure if anyone has a suggestion?

What do you mean by 'Active / Passive' as that isn't a default behavior of the product.   Do you have a basic Process Automation cluster setup and the load balancer is just only directing to one at a time or do you have two Domain level orchestrators reading from the same database?

If you have 2 Domain Orchestrators, that are not clustered, reading from the same database that both believe they are the only orchestrator they will both read the schedule, both attempt to launch the process and both attempt to run through and complete it.  

In the \pam\server\c2o\bin\c2osvcw.conf file what is set for each systems Duuid value shown here:

wrapper.java.additional.3=-Duuid=74f68a55-9d43-46d8-ad8a-626b66114fbc

In \pam\server\c2o\.config\OasisConfig.properties, what is set for

oasis.server.isCluster ?

If the Duuid is the same, but the Orchestrators are not setup with Cluster=true and both are up and running then each orchestrator will think it is the only orchestrator nd both will attempt to execute the scheduled process.

Hi Mike,

I believe John is referring to the DR feature we have provided in PAM 4.3 SP01. Please refer the below link for more details on the Active/Passive node setup.

How to Use the Disaster Recovery Plan - CA Process Automation - 4.3.1 - CA Technologies Documentation 

derekj: Can you please raise a support ticket so that our team can have a look at your deployment and suggest a possible solution or provide a code fix if required?

Thanks

Vinay

Derek - is this using the Brocade ServerIron ADX setup, or are you using the DR plan as outlined for CA Process Automation mentioned by Vinay?

If this is the Brocade ServerIron setup, CA Support most likely will not be able to assist.

We are using the Brocade ServerIron ADX configuration and not the DR plan.  We'll perhaps look at turning off the Process Automation service on one of the two servers, effectively breaking the redundancy and introducing a manual process to evoke a failover.

Ok, then my post above is still valid, Process Automation works schedules from the database itself.   If there are 2 active Domain Orchestrators reading from the same database they will both see there is a scheduled job to run and will both execute it as they have no way of knowing not too.    You will need to either look into the Active / Passive configuration Vinay linked, or as you suggested shut one of the Orchestrators down so only one is reading and writing to the database at a time.