I just finished upgrading a client over the weekend. We went from 10.1.500.3x to 10.2.500.13. I completely uninstalled Epicor (and task agent, Admin Console, etc), rebooted, and reinstalled.
After fixing the usual hiccups (things not tested in the TestUpgrade environment), I have a problem that I don’t know how to diagnose well.
The App Server’s memory usage simply continues to climb. It is slow, but I’ve had to recycle the App Pool 3 times today after it went from 2GB -> 24GB.
How can I see what is eating up the RAM?
I assume it’s a Server Process, Task, or BPM, but there are a lot of things to test and this is not happening in TestUpgrade (different machine).
You need to install and look at the Performance and Diagnostics Tool, take a memory dump, enable server logs, and analyze them with this tool. That’s all explained on that document.
Any chance there are some processess that run extra hard the first time? Like if a scheduled Generate PO Suggestions task is set for “Net Change”, the first time it would run would be like if Regenerative was selected. I assume MRP would be similar.
Well,
I feel a little sheepish. It seems to have resolved itself. It is now hovering at 3.2GB (yesterday it went to 23 GB 3 times and was giving errors).
I saw something similar a while back in about the same version of upgrade (from and to). We saw, while monitoring network traffic that two clients were generating a ton of requests. Turns out these people did not log out of the old version before the upgrade and seemed to be in some odd state of retrying the request. Once they logged out and upgraded the client the problem disappeared. Never really circled back to see if this was the root cause though. When you installed the new version and wiping the old was it the same server and application name?