What exactly is this process and how many of them should be running on a poller?
The Job Engine is used by the Universal Device Poller, APM, and EOC to do recurring work. It comprises the Job Scheduler service (central, dispatches work to local or remote job engines), the Job Engine service (one per server), and various Job Engine Worker processes. This separation facilitates scalability and robustness.
The number of Job Engine Worker processes will vary depending on how much polling work these pieces are doing. Lots of UnDP pollers or APM monitors will make it work harder, which may cause it to decide to spin up another worker process.
More than a handful might be cause for concern. How many are you seeing?
I am seeing 2, however I do have quite a bit of UnDP routines set up.
Example, pulling down the HP server info. CPU types, and speeds, and cores per.. drives and thier status and chassis status variables... each of which i tell NOT to keep historical data on and they are pretty much non-changing and dont need graphing.
I also do the same for the DELL servers we have.
Additionally I do the WINDOWS domain/workgroup per server (Please add this OID into the Node Details window) hint hint.
Lastly im pulling down the table of IP addresses on each system (Yet another item I would like to see added since systems can have many ips associated to it and its a universal OID that works on all systems.)
We are getting ready to price out and build a new NMS environment, so im now at a pass between 2-Dual Core's or 2-Quad Core's for the polling engine as I want to really get some more like APM and Custom OID's as well as the typical monitoring.
And since im kinda hitting on this via this thread. If you have multiple polling engines, does the Universal Polling per device move to the poller thats assigned to manage the node or does it all stay on the poller you built the Universal Poller variable on?
Two SWJobEngineWorker.exe processes is normal.
And to answer your question, UnDP polling will run from whatever poller the node is assigned to.
How about memory usage of the SWJobEngineWorker.exe. Mine is getting above a Gigabit of RAM and it starts sending out spurios messages about APM stuff being unknown/down/up. If I shut/start the service it tends to go back to working ok for awhile. This did this in both 9.0 and now 9.1SP1.
