Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I have a windows 2003 server here which occasionally froze every couple of weeks.

Now it is happening every day. It has been hanging sometime during the night. How can I setup a running log of cpu/program/memory usage to see if there is a program or task which is causing it to freeze?

I thought it could have been the backup, but I can run a backup task during the day (writing to the tape drive) and it works without any issues.

Any thoughts or suggestions would be very appreciated.

share|improve this question

2 Answers 2

You use the term "freeze", which would imply that the machine is halting. In my experience, if you're not seeing a kernel STOP (a "blue screen") but rather are seeing something like, say, blank video or a static video display w/ no activity (and, in particular, no mouse pointer movement) then I'd be more suspicious of flaky hardware than of software.

If the machine doesn't have ECC memory I'd bring down a copy of MemTest86+ and give the RAM a thorough test.

If it does have ECC RAM (and you're not logging any memory errors) or it passes a few hours of MemTest86+ w/o error (for non-ECC RAM) I'd consider giving the box a few runs of the manufacturer's own diagnostics.

There's certainly a chance that it's not a hardware problem, but "freezing" in Windows, in my experience, has usually been hardware-related.

share|improve this answer
    
The screen turns itself off after a certain amount of time. When I get back in the morning, no keyboard or mouse moments can wake it up. I've ran suppliers diagnostics on memory and it returned all fine. For all 6 sticks of RAM they all had: Multi-Bit Status=0,Single-Bit Threshold Count=0, Multi-Bit Status=No DIMM errors detected,Single-Bit Threshold Exceeded Status=No DIMM errors detected,Memory type=DDR2 SDRAM FB-DIMM –  Mike Dec 7 '10 at 4:32
    
I haven't run a memtest on it, it is slightly more difficult to do this because its a production server without a slave –  Mike Dec 7 '10 at 4:37
    
When you say it cant be woken up, I assume at this point all communication with the server halts also? –  JamesK Dec 7 '10 at 10:30

I had this same problem with one of my servers a few years ago. I wouldn't get any error logs indicating a problem. Event viewer was clean until I manually rebooted in the morning and it said "Previous shutdown was unexpected."

I swapped the Hard Drive, tested and swapped the RAM, swapped network cards. Nothing worked. Finally I took the original hard drive, put it into an identical server, and the crashes stopped. I could never replicate the problem, and every test came back clean.

I never figured out what the actual problem was, but the only options left were the motherboard and I guess the on board video card.

This is most likely a crucial hardware failure. Do you have a spare server you can failover to for a little while to test your hardware?

share|improve this answer
    
That sounds exactly like the problem I have. I wish I had another server to use as failover for a while, however the other two servers are in production as well. The event viewer is clean also. Ugh :( –  Mike Dec 7 '10 at 5:57
    
If you have bad hardware, you will eventually need to replace it. If the problem is justifying/proving this to someone else, maybe you can setup a regular desktop to run your application temporarily. Just to keep it running while you investigate the primary hardware issues, and then replace it if needed. –  minamhere Dec 7 '10 at 12:40
    
Do you have a warranty? Maybe this is much simpler. –  minamhere Dec 7 '10 at 12:41
    
Also, here was my original question on ServerFault. Maybe there are a few suggestions you can try that could at least lead you closer to identifying the problem. serverfault.com/questions/46511/… –  minamhere Dec 7 '10 at 12:45

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.