Question
TCS
IN
Last activity: 28 Mar 2017 9:37 EDT
Unused Default Agents in Production
We have multinode cluster as 2 JVMs pointing to 1 DB. We want to disable unused default agents in the production system. Is it required to enable default agents in all the nodes of cluster?
Capgemini Sverige AB
SE
Hi Srinivas,
why is that requirement to disable the default agents? are you sure that those agents are not processing anything with your application?
TCS
IN
Hi Srini,
There are many of the default agents are running now and some of the agents are not required in the production systems. We have identified the following in pega administration course.
GenerateStartingFlows - Relanted to Unit Testing
PropertyOptimization - Property optimization
PostAction - PMF Related
CreatePMFItem - PMF Related
UpdatePMFItem - PMF Related
Trying to check is there any other agents like this.
Pegasystems Inc.
IN
What do you exactly mean when you say unused agents? Which agents in particular are you talking about that you want to disable in production ?
TCS
IN
Hi Srini,
There are many of the default agents are running now and some of the agents are not required in the production systems. We have identified the following in pega administration course.
GenerateStartingFlows - Relanted to Unit Testing
PropertyOptimization - Property optimization
PostAction - PMF Related
CreatePMFItem - PMF Related
UpdatePMFItem - PMF Related
Trying to check is there any other agents like this.
And my intentions is to know, do we need to enable the default agents in both the nodes? Or we can run the agents on only one node as we run the our own agents in one node?
IN
Hi Srini,
In response to query "And my intentions is to know, do we need to enable the default agents in both the nodes? Or we can run the agents on only one node as we run the our own agents in one node?"
Ideally in multi node environment connected to LB, it's advised to have the similar configuration across all nodes. Consider the situation wherein the only one configured node is down (may be any "x" reason) and the need to run the agents are on priority, how would you fulfil this on priority specially in production emergency situation?
Pegasystems Inc.
GB
No it is not necessary to have all default agents running in Production and ideally you should not have.
Also in a multinode environment with batch processing its highly likely you will have two types of nodes, those handling end user requests which will be load balanced, but there could be another n nodes running batch processes via agents. You would only have the batch agents running on designted batch nodes not all nodes.
In some instances Agents can only run on one node, this may be due to design if custom. Although not ideal we have this with ADM in the decisioning area.
OOTB agents should be altered in different environments, what you have running in Dev is not the same as what you want in Test or Production.
Also the schedules should be altered as they are too frequent as provided OOTB and depending on your applications and end user base may need to be altered for when they run so that they do not impact end users.
considering a production system the following agents may need to be switched off (disabled so they dont appear in SMA or AES)
GenerateStartingflows not in production
those in PegaAESRemote if your not using AES
PropertyOptimisation yes in dev not in upper and prod environments
Refactor Copy/Move/Merge yes in dev not suitable in prod and higher environments
UpdateActiveNonQuiesceAdmnRequestors remove if not using HA clustering
No it is not necessary to have all default agents running in Production and ideally you should not have.
Also in a multinode environment with batch processing its highly likely you will have two types of nodes, those handling end user requests which will be load balanced, but there could be another n nodes running batch processes via agents. You would only have the batch agents running on designted batch nodes not all nodes.
In some instances Agents can only run on one node, this may be due to design if custom. Although not ideal we have this with ADM in the decisioning area.
OOTB agents should be altered in different environments, what you have running in Dev is not the same as what you want in Test or Production.
Also the schedules should be altered as they are too frequent as provided OOTB and depending on your applications and end user base may need to be altered for when they run so that they do not impact end users.
considering a production system the following agents may need to be switched off (disabled so they dont appear in SMA or AES)
GenerateStartingflows not in production
those in PegaAESRemote if your not using AES
PropertyOptimisation yes in dev not in upper and prod environments
Refactor Copy/Move/Merge yes in dev not suitable in prod and higher environments
UpdateActiveNonQuiesceAdmnRequestors remove if not using HA clustering
PMF items remove if PMF is not used
Decisioning items remove if decisioning is not used
pycleanUpSSOn if your not using get rid!
dare say theres a few othes too, you need to determine what your using where