This is just information for others who might run into this problem. It took weeks to discover the fault, so I figure I would help avoid a headache for someone else.
I've been having some odd issues with a server that runs Manager (continuously) for one of our offices. After about 5 days, the computer basically becomes unresponsive. Many services stop and I'm unable to remote desktop or VNC into it. A local user has to restart the server. After a reboot, it's good for another 5 days. I have checked the logs. The System event log shows event ID: 2019 with a description saying the non-paged pool is empty. It's the only server at that office, so it is setup as a domain controller. It has 1GB of memory. The server only does Wins, DNS and DHCP for that office. It doesn't host any applications.
According to some sheets from Microsoft, the error almost always means an application has a memory leak. They gave a great tip on finding the offending application. Run Task Manager and navigate to the Processes tab. Go to View and Select Columns. Add Handle Count to the checked list. Sort the processes by handle count. Per Microsoft, any process that is using more than 5000 handles most likely has a memory leak. The handle count should remain pretty consistent over many days. If it keeps growing, there is a problem. What is a memory leak? It's when a process does not release memory back to the pool when it finishes. In the case of Manager, it starts out under 1000. After 3 days it gets up to 14,000. At some point, it takes up all available non-paged memory. That's when the server becomes unresponsive. I've seen it as high as 56,000.
I have a second IP Office system running Manager. It has more physical memory and is not a domain controller, so it takes much longer to see a problem. I'm sure that having the problem server running AD adds to the problem or at least speeds up the symptoms.
I've also found VM Pro to have some similar memory leak issues, but not nearly as critical. If you reboot your server on a regular basis or don't keep Manager running, you'll never see this problem. But if you do see it, at least now you'll know why.
What's the solution? Reboot every few days, install more memory or install a newer version of Manager that possibly does not have this issue.
I hope this helps someone out.
Tom
I've been having some odd issues with a server that runs Manager (continuously) for one of our offices. After about 5 days, the computer basically becomes unresponsive. Many services stop and I'm unable to remote desktop or VNC into it. A local user has to restart the server. After a reboot, it's good for another 5 days. I have checked the logs. The System event log shows event ID: 2019 with a description saying the non-paged pool is empty. It's the only server at that office, so it is setup as a domain controller. It has 1GB of memory. The server only does Wins, DNS and DHCP for that office. It doesn't host any applications.
According to some sheets from Microsoft, the error almost always means an application has a memory leak. They gave a great tip on finding the offending application. Run Task Manager and navigate to the Processes tab. Go to View and Select Columns. Add Handle Count to the checked list. Sort the processes by handle count. Per Microsoft, any process that is using more than 5000 handles most likely has a memory leak. The handle count should remain pretty consistent over many days. If it keeps growing, there is a problem. What is a memory leak? It's when a process does not release memory back to the pool when it finishes. In the case of Manager, it starts out under 1000. After 3 days it gets up to 14,000. At some point, it takes up all available non-paged memory. That's when the server becomes unresponsive. I've seen it as high as 56,000.
I have a second IP Office system running Manager. It has more physical memory and is not a domain controller, so it takes much longer to see a problem. I'm sure that having the problem server running AD adds to the problem or at least speeds up the symptoms.
I've also found VM Pro to have some similar memory leak issues, but not nearly as critical. If you reboot your server on a regular basis or don't keep Manager running, you'll never see this problem. But if you do see it, at least now you'll know why.
What's the solution? Reboot every few days, install more memory or install a newer version of Manager that possibly does not have this issue.
I hope this helps someone out.
Tom