PE,TaskMgr not running on Windows 2003;EventID 333

OK - so, to be sure, this statement is not actually the case?

let me see on that. Since we are using virtual machines. They must not be very different.

Thanks for the clarification.  If /basevideo has no impact, I'd (as mentioned) be curious as to the effect of limiting the amount of RAM.  There seems to be plenty free, so I do not expect that doing so will have any impact on performance...

...After the /basevideo experiment, correct? (Just to be sure - only want to be changing one thing at a time...)

Hi all,
After more than 3weeks troubleshooting slow terminal servers the war seems to have been won. It would seem on the original problem server , the video card drivers were the culprit. They were using a high number of sysptes resulting in the shortage PTEs  and the server would literally grind to a halt after some time. This is the second week and the server seems behaved now handling more connections than it was normally would. When the problem started the server only handled less than 10 connections before it came down to snails pace. Now its taking btwn 30 and 40 concurrent connections.

The second server which also exhibited just about the same symptoms , slows down till it grinds to a halt. Taking less than 10 connections as well. But on this one there were no shortages on PTes or memory. The only unusual thing that I realised what was handle count for My-Sql- service file running on the system which seemed to be leaking handles somehow. I started monitoring it last week when Motolov said the fact that you have the problem does not mean its the same issue.

Thanks motolov I can at least do something about the complex problems that I meet everyday. 

I have been working well until yesterday. Apparently this time around the issue is slightly different but related to the original problem. Server hang. Apparently the server is working but now sometimes it drags since yesterday. I have checked out the pool allocations sysptes, overall memory usage which all seems with reasonable limits. Live debug shows that there has been 1 syspte allocation failure, 30 pool allocation failures which I suppose happened may be yesterday. At this point in time, everyone logged on is working fine but I cannot run perfmon either locally or remotely. Context switches -16 225 000 000.00 with cswitch delta - 35000 for system idle process.interrupts have 5 500 000 000.00 and cswitch delta of 2700, DPC has 1 500 000 000 and cswitch delta of 2100. What else can we look at. And Remember this 16 Proc Server,with 36gb Ram.


Edited by zizebra - 30 June 2009 at 11:50am