Question
How to configure the Pega-ProCom agents to run on specific node in multinode environment ?
The PRPC 7.1.5 is hosted in our In-House cloud environment, were we have multiple node (tomcat's) with single rule base.
Currently by default OOTB Pega-ProCom agents are running in all nodes, but we have requirement to run few agents only on some nodes. For example the PDM is connected to single node and we want the SendCorr agent to run only on that particular node to print the Mail(Data-Corr-Letter). we don't want to configure the PDM in all nodes due to maintenance and scalable issues. Please suggest.
***Updated by moderator: Marissa to close post***
This post has been archived for educational purposes. Contents and links will no longer be updated. If you have the same/similar question, please write a new post.
You can use "Agent Schedule" records for specific Agents viz. Pega-ProCom Ruleset and enable or disable them on specific nodes. The tricky part is that Agent Schedule record is unique to a NodeID (viz. 6d2097879de17b6e6f08201929312d86) and you will need to ensure this NodeID does not change upon successive restarts. Making sure each node is configured to use its own unique PRPCTemp directory, you won't have an issue with NodeID change.
Thanks. But that's not really manageable. It is not really practical to rely on ID's that are random MD5 hashes. This is why we were leaning towards making a custom setting that we could control.
If we have configured the Temp directory correctly for each node , The below code will give the exact path of temp directory for the current node. This is used in activity syswebinfo which gives us the Environment detail . We can use this in first step of all agent and proceed it if it is running in the correct node else we can exit the activity.
explicitTmpDir = tools.getSystemSettings().getFSSetting("initialization/explicittempdir", "" , true, false);
re: "If we have configured the Temp directory correctly for each node..."
What do you mean by "correctly? Should it be configured uniquely for each node? And remember, our goal in processful, decoupled architecture, is not to unnecessarily bind unnecessary things like the node name and the agents it's running. That's why I proposed a new setting.
And what do you mean "first step of all agent" -- really? What about the OOTB agents that we can't override?
Yes Jon I agree, This works only for custom agents and not for OOTB agents.
Glad you asked. Here's the solution I posted internally last week. I haven't actually coded it yet, but my sense is that this should work. Curious what the community has to say here.
[Moved this to a separate document for editability: AgentManager - How to programatically manage bulk init/start/stop of Agents - July 3rd 2015]