J
Jesse Pastrano
Guest
Hi there,
we are facing the following problem:
On our Windows Server 2003 SP2 Standard Edition File server we are forced to
reboot due to low Paged Physical Kernel Memory. Once we reach around 350,000
K, the server stops responding to client requests and we can not access it
from the console.
(Eventid 2020: The server was unable to allocate from the system paged
pool because the pool was empty.)
This is a Fileserver with approx 2 TB of data containing AutoCAD files and
user home directories.
All volumes use Volume Shadow Copies for Folders to give our users
the Previous Versions possibility. The target for these Shadow Copies
is another Volume from the source Volume.
During the day we have between 2000 and 2500 open files at one time and 250
server sessions.
Also noted from perf mon, the Current Disk Queue Length can max around 94
and avg about 6.
We already implemented the fix mentioned in
http://support.microsoft.com/kb/312362/en-us
(PoolUsageMaximum is already down to 40)
and also
http://support.microsoft.com/kb/317249
To clear some paged memory it helps to create a shadow copy manually,
but this only helps for an hour.
The source and target volumes for the Shadow Copies have very low
fragmentation.
Any help or ideas how to solve this would be very appreciated.
Thanks and kind regards,
we are facing the following problem:
On our Windows Server 2003 SP2 Standard Edition File server we are forced to
reboot due to low Paged Physical Kernel Memory. Once we reach around 350,000
K, the server stops responding to client requests and we can not access it
from the console.
(Eventid 2020: The server was unable to allocate from the system paged
pool because the pool was empty.)
This is a Fileserver with approx 2 TB of data containing AutoCAD files and
user home directories.
All volumes use Volume Shadow Copies for Folders to give our users
the Previous Versions possibility. The target for these Shadow Copies
is another Volume from the source Volume.
During the day we have between 2000 and 2500 open files at one time and 250
server sessions.
Also noted from perf mon, the Current Disk Queue Length can max around 94
and avg about 6.
We already implemented the fix mentioned in
http://support.microsoft.com/kb/312362/en-us
(PoolUsageMaximum is already down to 40)
and also
http://support.microsoft.com/kb/317249
To clear some paged memory it helps to create a shadow copy manually,
but this only helps for an hour.
The source and target volumes for the Shadow Copies have very low
fragmentation.
Any help or ideas how to solve this would be very appreciated.
Thanks and kind regards,