Concerned about high CPU usage

[wahnula]

Hello,

My SBS2k3SP1 32-bit home-brew server has been acting up lately. It's a dual-Opteron 1.6GH, RAID 1 for OS (onboard nForce4 controller), RAID 5 (3Ware Escalade card) for data & Exchange store (all SATA Raptors), 4GB ECC RAM.

Starting with a power outage Wednesday night during SBSBackup, the machine was safely shut down by the UPS software. However, the next day I saw a flurry of NVRAID errors (RAID 1 controller) and received warnings of low System Idle Process values, indicating high CPU usage. In Task Manager, one of the CPUs is at the top of the graph, the other near the bottom. The HDD activity for the OS array was glowing deep red.

Upon reboot I saw there was a degraded RAID 1 array, instead of just replacing the drive (I had a spare too ) I rebuilt the array and continued to boot. It spun for 15 minutes on the splash screen, then stopped and said something about APIC being needed for a 64-bit OS (?), basically saying "you're hosed". I tried booting off the good drive, replaced the bad one but too late. Luckily I had a good backup from earlier that day.

Long story short, recovered just fine from backup, but I still have the pegged CPU, one more so than the other. One stays at or near 100%, the other one goes between 20-80%. System Idle says 97% but CPU graph in Task Manager says lots more, and I keep getting notices about low SIP. No NVRAID errors and no glowing HDD light though.

We have been without Internet for quite a while, it's been intermittent the last few weeks (thanks Ike) and is off now. I noticed attempting to open a browser really pegged the second CPU. There are no errors in Event Viewer except a bunch of W32Time errors due to the Internet being down, which will be fixed (again) today. Anything jumping out at anybody? I'll take all ideas. Windows Auto-Updates is disabled. A/V on gateway, server and clients, but I have not run any malware scans on the server as I've no Internet to update. Thanks for your thoughts.

Tony

Users helping Users...

 

[BadBigBen]

For test purposes, try to turn off the onboard NICs and unplug the cables... I've seen similar things due to a network problem, probably a spike over the ethernet cables...

and even though your UPS turned off the machine, it may not have cought the spike due to the power outage...

Ben

"If it works don't fix it! If it doesn't use a sledgehammer..."

 

[wahnula]

Thanks for the reply Ben. There's still work going on here so I can't disable the LAN (Quickbooks and prospect scheduling) but I did try disabling the 2nd NIC that goes to the outside world...no perceptible change.

Tony

Users helping Users...

 

[wahnula]

UPDATE:

Cable guy put the line back up. System still wonky. I went to connect the external backup drive (File copy & Exchange Store back-back up, I have an internal IDE for SBS Backup. When I connected the drive, that was the last straw. The server locked up and the screen went black. I went to my workstation to RDP to the server & reboot, halfway through I got a runtime error SO...the server is still chugging away with no way to get at it.

I memorized the sequence to shut down with the keyboard, it worked! Upon reboot, the NVRAID was degraded, I rebuilt it and have no access to Exchange, apparently the logs & database detached. Time for a quick post to Exchange 2003 forum to get a crash course on ESEUTIL. Still erratic and heavy work on the CPUs.

Tony

Users helping Users...

 

[BadBigBen]

Have you tried another PSU?

other than that I am at a loss at this stage...

Ben

"If it works don't fix it! If it doesn't use a sledgehammer..."