AS2K - Crashing Under Stress! 1

[tylan]

Specs:
Windows 2000 SP4 (Fresh install to eliminate W2K as Prob!)
512MB PIII 600 or 700
AS2000, Tape Lib Option, Open File Agen ---> SP4, q020824, q035241, q041394.
Overland SDLT Loaded Xpress
3Com 3c2000-T (32bit GB NIC)
Backup Other servers via NT/2K Client Agent
Backing up ~25GB over network.

Server started BSOD a month or more ago. I feel it started around the time that W2K SP4 was applied - but can't prove. Server seems to BSOD on ntoskrnl w/ 0x0a or 0x1d. If it doesn't BSOD it basically drops off the network. Can't ping it, can't ping FROM it. Server seems to crash only when it backs up the 2nd w2k server (client agent). Remote server happens to have a GB Nic also. Seems related to high x-fer rates. I've spoked w/ 3com and they had me bump up Rec & Send Buffers etc (advanced nic settings). But to top it all off, it seems random-ish. Tonight it crashed on attempt 1 and 2, but seems to have made it to the verify on attemp 3 -- I haven't changed anything! I've checked processor fan, IRQs, IOs, and Memory addresses for the NIC, and they all seem fine to me.

That's the story since the fresh reload of the W2K Backup Server. If anyone has an idea I'd GREATLY appreciate help!

Thanks In Advance!
Tylan

 

[davidmichel]

Make sure the NIC does not share IRQ with anything else.

It does sound like it is related to the NIC.
Have you installed the latest driver from 3Com?

Here are two idea.
1) Replace the 3Com with a different NIC, best if it is a different brand. Uninstall, restart, then reinstall the driver.

2) Trobleshoot the Blue Screen message. A review of the first four lines of the Blue Screen might help to identify it. If it can be identified as relating to the 3Com then it is their issue and they need to take ownership.

Suggestion 1 will probably be your quickest solution, and based off what your time is worth it might also be the cheapest.

 

[tylan]

I think the problem I have now replaced a problem I had before. The 3C2000 NIC is relativly new.

Talked w/ 3com for quite a while today. They feel that the problem is linked to using thier card which is PCI 2.2 in my Super Micro Board which is PCI 2.1.

Any Ideas? Is that a fact?

 

[davidmichel]

Can not confirm but it makes sense to me.

How well can the PCI 2.1 bus handle PCI 2.2 function calls and higher performance. I would have thought that a 2.2 version adapter would be fully backward compatible with 2.1 by disabling the newer features but if that is what they are saying.

 

[tylan]

I was down with the "backwards compatible" thing. I'm gonna see what I can find out.

 

[tylan]

OK, So its been a while....

I've tried everything and this server just BSODing.... rather randomly. I even got MS in on this, we opened a case under the free 'problem was caused by a service pack' provision. MS looked at the dumps and insisted it was the NIC. I told them we were on our 3rd NIC already, but thanks anyway. I was getting desparate, I even called super micro about IRQ/Card Positions. They said the server was near perfect in its config. One tech suggested setting the BIOS so that the cards are AUTO, not forced IRQs. Another tech happened to mention something about 700MHz coppermines, and I told him I'd check. I kinda wrote him off cause he was focused on 700MHz, and I was pretty sure I had a 600MHz.

Well, after going around and around I found that the issue was related to my 600E MHz Coppermine. The Super Micro P6DBE (Dual Slot) was designed to use .25 micron processors, but the 600E is .18 micron. To make a long story short I needed a Active Terminator that didn't ship w/ the board or another 600E. I scarfed one that was in another server, and TADA no more BSOD! It even got rid of my W3963 Errors! I guess it was just a voltage issue all along!

 

[jtb]

Was there any special BIOS or driver pack needed to load W2K to that Mobo?

JTB
Have Certs, Will Travel
"A knight without armour in a [cyber] land."

 

[tylan]

No. Worked fine.