Hi Pixie,
This could be a memory leak caused by one of the dllhost.exe processes on your polling engine server. To check this, open up Task Manager on it and sort by "Mem Usage." If you see dllhost.exe as one of your top uses of memory (mine usually use 25 MB while this one process was using 192 MB), try the following configuration changes on it:
Start – Programs – Administrative Tools – Component Services – Computers – My Computer - COM+ Applications Right click "Solar Winds Orion NPM" – Properties – "Pooling and Recycling" tab Change "Pool Size" from 1 to 5 Change "Lifetime Limit" (minutes) from 0 to 5
Make sure "Expiration Timeout(minutes)" is set to 15
Reboot the server
I hope that this helps. Best of luck.
Steve
I was also having this when I was having SQL performance issues. I resolved by enabling AWE memory allocation on the SQL server.
Did changing the COM+ applications properties correct the issue?
I beleive I have the same issue but he poller and db are on the same server so I am sure there are no firewall realted issues.
Hi Clint,
I discovered that it corrected the problem where our primary poller was failing over to our secondary one because it couldn't open a connection to the remote SQL box. This is when we noticed the error message in the Event Viewer on the polling device. This happened usually four or five times a day or night at random times. However, we're still seeing those same messages in the Event Viewer (no such messages seen on the SQL box). I asked Solar Winds if the changes made were a permanent fix or simply a band-aid. According to them, they're working on a service pack or hot fix to stop this message from showing up in Event Viewer (please see below).
"Thank you for the response. This is permanent fix for the dllhost.exe. As for the event error message, this is due to the change setting, and is being looked at for a service pack to correct this. Thanks."
Sincerely,