Hi Michael,
Thanks for verifying that from your LAB setup. I suspect an issue with the F5 load balancer configuration. I'll have to spend more time with the network guru.
Please correct me if my understanding is wrong:
In simplified communications, the Agent maintains a connection with the Orchestrators (through the load balancer, port 443) to be informed when there is work to do. (I assume this as the Agent is not listening on any port and I see a network connection between the Agent and the load balancer. Viewed using Process Explorer.)
Here is the issue I am seeing:
- I log directly into one Orchestrator directly (without going through the load balancer).
- I log into the other Orchestrator directly (without going through the load balancer).
- I launch the Agent Console on the first Orchestrator and try to get the Status of the Agent.
- I launch the Agent Console on the second Orchestrator and try to get the Status of the Agent.
- I find that if I wait a while (~15 minutes) and try the second Orchestrator Agent Console to get the Status then it will show Active but the first Orchestrator will now show Inactive.
I wonder if I should see two connections between the Agent and the load balancer, one for each Orchestrator.
This also manifests itself with Process Operators assigned to the Agent sometimes failing with System Error "Message could not be posted to the node."
Cheers,
Lindsay