Windows Forum

Ask something
Discussion Locked

Software SCSI RAID problems

0 helpful
5 replies

by cocosk

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.
I'm running NT4 sp6a using an Adaptec 29160 SCSI adapter to host 2 NT software RAID 0. Each RAID consists of 4 drives in an external case. The RAID boxes are daisy chained using Ultra SCSI cables. The problem is drives are going off-line. All drives are 18GB IBM SCSI LVD drives. I've updated the firmware for the IBM SCSI drives. When attempting to capture video to either RAID, usually a drive will stick and I get the error "system process, lost / delayed write data, the system was attempting to transfer file data from buffers to device \hardisk2\partition1\$Mft"". This error occurs randomly on any of the 8 drives. In the event viewer, I get event id 9 & 11. According to Microsoft TechNotes, this could be a SCSI cabling, termination, failed device, or bad drivers. I've installed all the latest from IBM for the hard drives. I have the latest Ultra160 drivers for the 29160 card. BIOS is disabled on the 29160 because I have no boot device attached. I do have a physical LVD/SE terminator at the end of my external SCSI chain. I've upgrading the cable that attaches the 2 RAID boxes. I'm still getting the errors. Some things I suspect, but can't be certain are the 29160 card is going bad, maybe exceeding cable length, or IBM hard drives are crap. I thought cable length for LVD was 15 devices at 32 feet. My cable from CPU case to 1st RAID is 3 ft. Cable from 1st RAID to 2nd RAID is 3 ft. Add in internal cabling & I get close to 32ft, but don't think I exceed it. Are there any good troubleshooting or testing of the Adaptec 29160 card? Is there some registry hack for a timeout period? Any other troubleshooting methods or suggestions? Thanks!

5 replies

by boyhowdy

There is a good chance that including all of your bus cabeling and backplane circuitry inside the boxes, you are exceding 32 feet.

I have always used only one RAID set per channel to be extra safe on not eceding the cable length.

Also, what kind of server are you running? I've had problems in the past with running Adaptec cards in Compaqs and HP's. If the server manufacturer makes a RAID controller, you might want to consider going with that one. The benifits of getting a card that is designed for your server by the manufacturer far out weighs the price difference.

As far as trouble shooting the card, I'm pretty sure there is a troubleshooter in the CI/O software that is included with the drivers, and I've never had a problem with Adaptec tech support. Just in case, you might want to have a good session of RTFM. If you've lost the manuals you can get them in .pdf format from the Adaptec site.

Let me know if this was helpful.

BoyHowdy

by cocosk

Thanks for your suggestions. I've done more troubleshooting with no definite answer. I'm sure the problem is not cable length as I'm down to one RAID (4 drives) on this one chain and still getting the errors. I need to buy all new components and start replacing them one at a time..... when I find some cash to do so. I'm thinking it could be the Adaptec SCSI card, a CPU heat issue, or cabling somewhere between the CPU & the drives. Thanks again!

by cocosk

Thanks for the response. This is not a server, just a PC clone workstation in a simple configuration. I thought about buying a RAID controller, but decided it's not worth it since I'm only doing RAID 0. Can't afford a backup strategy yet. Weird thing is that this configuration has worked previously. I also remembered that I received the same error when I only had the 1st RAID attached, so I don't think cabling length is really the issue. I am beginning to suspect heat or processor. CurrentlyI have Dual PIII 500Mhz with 512Mg RAM. If I leave the system running for an hour or so and try to access the RAIDs, I get bad lag & the error. Today I noticed the CPU idle time was at 99 & I had major spikes on both processors. When attempting to shutdown I got a BSOD citing the ntoskrnl.exe. I'm beginning to think a flush & fill would be best, although I'm not positive it'll solve the problem. After the BSOD, I opened the case & discovered my processors were really hot. This case (not put together by me) has really bad design. THe power supply is right near the processors and the heat between them was really high. I decided a better case with more ventilation was needed, so I bought one & moved everything over. I'm not testing to see if after an hour of no use on the computer, I still get the errors. I'm thinking that since I could access data fine at initial bootup & if the computer sat on for a while, it was heating up to where the processors were starting to choke. I also remember getting this error (Lost/Delayed write data) on my zip, which isn't anywhere near the SCSI chain. This leads me to believe that it's not a SCSI issue or cabling, but maybe processor heat, bad RAM, or just corrupt OS files. Any more ideas? What do you think? Thanks again for the response.

This discussion is closed to new replies.