Need Info on-Best Practice on what agent configuration in multi host/muti node environment for Pega 7
Here is the scenario. There are ~ 50 agents OOTB in pega 7.1.7/7.18. We multi node environment in production ( 64 jvms). We want to know basic answers to questions like
Do we need to run this agent on all JVMs(ex: System pulse) or it can be enabled only on one or few nodes(Fr example: if it’s a maintenance agent for doing housekeeping at DB side, we don’t need it on all JVMs).
most of the agents that come OOTB should not run on ANY node in production --
- many agents implement features that are not used in most Pega7 appplications (like updateAdaptiveModels, ProcessConnectQueue and many others)
- many agents implement features entirely inappropriate for production (rule refactor, updatePMF, runTestSuites, etc).
Agent management is a bit of a horror show for most customers as we ship final rules that enable tons of agents by default and its a true pain on a very large system to go node by node and turn off or tune the agent schedule. Some folks in the field just might have cracked some rule agent queue records to override default behaviors and make life easier for customers -- but those rules have to be re-done for every ml that might add a new agent.
Attached is a recent recommendation I sent to a customer.
Do note that there is an AES dependency version on the PegaAESRemote agents - turn all on with AES 71 and PDC; disable 3 of 4 with AES 3.5; turn all off if you dont have AES
If this is 717+ be sure the DNode listener is turned off ...