Topic: server locking up

[m alden]

I have a new 4.1.1 e-smith box with a 433 celeron, 128 ram, dual 9 gig seagates (in software RAID 1) on an Adaptec 2940U2W scsi controller that I have been testing before putting it into active service.

It runs fine with no load.  

Problem is when I start to do heavy file transfers to it.  After about 4 or 5 minutes, I get a complete system lock up.  No response from the box at all.  I have tried all different combinations with the drives:  different SCSI card, different cables, different drives, no RAID etc.  Always the same.  Even if I install on one drive with no RAID.  Same thing.

I tried running TOP while I was doing some file transfers, everything seems normal until slowly the files stop transferring, and TOP slowly stops refreshing, and then it stops completely.  I then cannot access the machine at all.  I have to hot boot it, and then usually only running fsck will bring the system back up.

Any ideas?
Thanks

[Charlie Brady]

m alden wrote:

> Problem is when I start to do heavy file transfers to it.
> After about 4 or 5 minutes, I get a complete system lock up.
> No response from the box at all.  I have tried all different
> combinations with the drives:  different SCSI card, different
> cables, different drives, no RAID etc.

Have you tried a different NIC? Sounds like a NIC hardware and/or driver problem.

Charlie

[m alden]

Charlie,

Thanks for the reply.  The box has two different NIC's in it.  Both new.  One an Intel 10/100 and the other a Kingston on the Tulip driver.  I have switched back and forth between the two NIC's multiple times as I thought the same as you.  I will continue to try down that avenue with completly different NIC's.

Anything else that I can look at?

m alden

[m alden]

Charlie, I just tried a 3COM 3C905TX and the same problem occured.

Ideas?

Thanks!

[Craig Foster]

Is this with the Tulip installed still?

Call me stubborn, but I still use the de4x5 driver with tulips, ever since the timeouts from about a one and a half years ago. I'd recommend the Intel 10/100 cards any day, but I still wouldn't rule them out being faulty

I'd also see whether the IRQ's are right. Boot the machine and put an empty floppy in the drive. It should pause (saying that it's not a boot floppy) showing what the computer thinks the IRQs should be , not ACPI or the OS. Possibility is to also reset the ESCD Data or PNP OS to no, in the CMOS.

If you've covered all this, then you are quite entitled to ignore me entirely

Craig Foster.

[m alden]

Just in case anyone ever has this problem:  I changed the motherboard and the system no longer locks up.

Go figure...

[m alden]

Just in case anyone ever has this problem:  I changed the motherboard and the system no longer locks up.

Go figure...

[Bruce]

Not the answer I wanted to hear.  I'm having the same problem.  It first showed up when I tried to do a backup to the workstation.

[Bruce]

It appears to have fixed the problem. Now I have to find a home for my ghz motherboard.

[m alden]

Really funny how that works...  That board that was giving me trouble works fine under Win2k & 98.  I have not tried a 'regular' version of RH7, but I will soon.  Interesting to see what happens.

Sorry about your gig board.  I guess it will be a while before there is true compatibilty.  Too many manufacturers.