asked on

Kernel Paged Pool and Kernel Non-Paged Pool

Experts,

I'm attempting to diagnose what I suspect is a memory leak on windows server 2008 R2. I'm closely monitoring the "Paged" and "Non-Paged" memory pools.

How can I figure out what the quota is set to for these pools? Essentially this is what happens:

1. Disk I/O drops to 0
2. If you RDP into the server it will let the user in but explorer.exe crashes every 5 seconds
3. 30 Minutes later the server will stop responding completely

Log files show no specific error messages. HP Insight Manager shows all hardware healthy. I can only imagine that there is a memory leak to cause these types of server behaviors.

If I can determine what the quota is set to for those pools then I might be able to find the culprit.

Thanks,

Brent

BWXT

Can you provide some spec's on the system in question, also do you see anything in event logs relating to paged-pool memory out of resources or the like? This can be a very technical thing to troubleshoot, so first we'd need to make sure this is definitely your issue. Have any changes been made recently to the system, the user base increased drastically, or anything else we should know? (Software installed?)

As far as quota's, Windows dynamically chooses the max size of the pools and allows a given pool to grow from its intial size to certain maximums....

64 bit system Nonpaged pool can get up to 75% of physical memory or 128GB, whichever is smaller. Paged pool max is 128GB.

32 bit systems. NPP, 75% or 2GB, whichever is smaller. Paged 2GB.

To troubleshoot a memory leak the first thing you'll need to know how to do is use Process Explorer, Poolmon.exe, or Resource Monitor in 2008.

Read this first:
http://www.simple-talk.com/sysadmin/general/troubleshooting-nonpaged-and-paged-pool-errors-in-windows/

ASKER CERTIFIED SOLUTION

bdeback

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

bdeback

ASKER

chkdsk /f is what fixed this issue. We ran chkdsk /verifyonly which determined that filesystem check needed to be complete because it found errors.