@Dev.Pega007  After an upgrade, Pega regenerates agent schedules per node and many end up enabled, so agents run on every node. Start by fixing node types: only the servers meant for background work should have the BackgroundProcessing node type, and web nodes should not. Restart those nodes so the types take effect. Open each Agent rule and set it to run only on selected node types, usually BackgroundProcessing, rather than on all nodes. In Admin Studio, stop the agent on unwanted nodes, then disable or delete the Agent Schedule records for those nodes so they do not restart. If schedules look stale, delete the Data-Admin-Agent-Queue entries and restart to regenerate them with the new settings. Verify there are no duplicate schedules from old node IDs or cluster names and clean them up. Consider migrating legacy agents to Queue Processors and Job Schedulers which honor node types consistently. After changes, resave the agents and do a controlled restart to confirm they now run only where intended.