Hi. Has anyone experienced performance issues on OM using agent groups mode ALL? We had an agent group with every UNIX agent we had on the system and this caused us problems. It was over 800+ agents. We wanted to run a cleanup job against all UNIX servers. Thought this would be a good solution. Even with the max parallel option selected to limit the # of concurrent jobs - we still had performance problems.
Support told us that this was not meant to be used with this many agents. We changed our process back to doing a PREP_PROCESS_VAR through a variable with all 800+ agents listed and submitting the job that way. That works fine.
However, we're trying to determine what the tipping over point is. Has anyone had experience with this and could provide any insight as to how many agents we should limit this agent group mode to? 100? 200? Without doing extensive testing I'm trying to set a guideline for our users on what we'll be able to allow.