It is a new server ~month old: 2.4Ghz Intel on Intel board, Hyper-threaded, 1GB DDR, RAID 10 SATA150 on Promise FastTrak S150 TX4, 4000MB Virtual Memory (swapfile, pagefile).
The performance monitoring shows two other interesting items.
Page Faults and Transitions Faults running parallel with peaks of 1500 within every minute.
Just trying to find some reference to what is the normal scale I found SBS Health Monitor’s default for Page Faults > 200 to trigger a warning. Well I sure as hell got that beat!
My understanding of Faults: Transitions Faults is where it is the pagefile just in a different location than where it thought (originally) it was. Page Faults means it isn’t in memory at all.
To clarify on what I meant my parallel is that Page Faults is ~10% more than Transition Faults, following each other almost perfectly in the monitored view. Thus if Page Faults jumps up to 1500 then Transition Faults jump right along with it to 1400.
I think Page Faults includes Transitions Faults, so Actual (Hard) Page Faults are 90% less than what the monitor is showing. So now the question is why so many Transition Faults?
Is it normal for SQL to have peaks of 1500 Page Faults on a 50Mb, ~60k Record one to many (8 more tables) database server.
The good news is that I haven’t had a system crash, since I disabled Promise’s PAM services.
I wrote a script.vbs to restart SNMP (net stop service, net start service) and used scheduler it to run every 24 hours. Until I find out what is going on.
Do I need 2 GB or memory? Or what?
I’ll give a 1000 points to who ever gets this nightmare solved.
Via another 500pts Question just referring this one, if it doesn’t violate any EE rules.
Since it is running on a dedicated W2k SP4 Server
I also put it in Windows 2000 area,
Please see: http://www.experts-exchange.com/Operating_Systems/Win2000/Q_20802243.html