-------------------------------------------------------------------- NOV-SCS2.DOC -- 19980115 -- Email thread on NetWare and SCSI devices -------------------------------------------------------------------- Feel free to add or edit this document and then email it back to faq@jelyon.com Date: Mon, 27 Oct 1997 17:36:50 -0600 From: Joe Doupnik Subject: Re: Adaptec 2940U / Netserver 4LF >Your problem isn't in bios versions or anything like that. > >When you install two SCSI adapters, only ONE of them can be set with >boot set. What happens is that when the one controller boots, it won't >allow the other one to perform a boot to the disks at the same time or >you would get two systems running at the same time. > >If you look while the system loads, you will see that only one comes up >with the BIOS message, the other will be silent. > >You can't as far as I know, perform the operation you want, ergo, boot >automatically from the second SCSI if the primary fails (adapter or >disks), it simply goes against the laws of adapters. ------------- There is a way out that works. Use a dual channel SCSI adapter such as the Adaptec 3940. That will boot from either channel because a single Bios controls both. With two boards one must have its Bios turned off. Joe D. ------------------------------ Date: Fri, 31 Oct 1997 10:22:18 GMT From: Kens Mail List Host Subject: Re: Adaptec 2940U / Netserver 4LF >Your problem isn't in bios versions or anything like that. > >When you install two SCSI adapters, only ONE of them can be set with >boot set. What happens is that when the one controller boots, it won't >allow the other one to perform a boot to the disks at the same time or >you would get two systems running at the same time. > >If you look while the system loads, you will see that only one comes up >with the BIOS message, the other will be silent. > >You can't as far as I know, perform the operation you want, ergo, boot >automatically from the second SCSI if the primary fails (adapter or >disks), it simply goes against the laws of adapters. This is not true. We've configured a number of systems this way here. Adaptec SCSI cards work very well with multiple cards installed. You MUST, however, configure each adapter to have a different bios address. This way each bios will detect and support any dos drives that are connected. If the first adapter has no dos partitions then the second will become the C drive and will boot as normal. We have the following setup: +--------+ +--------+ | Card 1 | | Card 2 | +---+----+ +---+----+ | | | +--------+ | +--------+ +--+ Disk 1 | +--| Disk 3 | | +--------+ | +--------+ | +--------+ | +--------+ +--| Disk 2 | +--| Disk 4 | +--------+ +--------+ Disk 1 and 3 have a DOS partiton and a NW partition and Disk 2 and 4 have just a NW partition. At startup disk 1's dos partition gets assigned to drive C and disk 3's to drive D. If we simulate failure of disk C then the card doesn't detect it and the second card assigns drive C to disk 3. Dos then boots of disk 3. ------------------------------ Date: Fri, 5 Dec 1997 13:46:26 -0500 From: "Jeffrey A. Tinch" Subject: Re: SCSI Drivers Unloading w/Arcserve (LONG) >I have a Compaq 1500, P166, 128MB RAM, 3.12 Fully Patched. Where we were >at. I had one 4.2G SCSI drive, SCSI CD-ROM, Adic 8008 Dat drive all on the >embedded SCSI controller on the server. We were experiencing mysterious >freezes of the system, utilization would jump to 39% drop all connections >and would not down properly. It was not an abend just a freeze. I would >have to power off, but all would come back up normally. > >I also added another 64MB RAM, we were at 64MB, and problem as well did not >go away. This was in anticipation of adding another OS2 name space volume, >see below, no as a fix to our problem. >We had to add another drive a 9.1G SCSI again a compaq drive, this drive >could not be recognized by the embedded Compaq SCSI controller. I >installed and Adapted 2940, Switched the two drives to the Adaptec and left >the DAT and CD-ROM to be handled by the Compaq SCSI. It should be way more >efficient that way... I installed quite a few compaq servers in the past 2 years, We always had problems getting adaptec controllers to work in compaq servers while trying to use compaq internal scsi controllers. Even had one site that had only adaptec controllers in a prosignia that would lock after about 30 days of stable use, we put in compaq scsi and problems went away. We got to the point were we just used only compaq controllers in compaq servers (don't use their clone version of aha1530 though), After instituting this policy we no longer had to deal with these types errors. We also always put the cd-rom and Tape drives on the internal scsi if it had one. Or added seperate card for them. By using seperate controllers we no longer had to deal with call backs, because of lock ups caused by cdroms & tape that are a lot slower than HD and would tend to screw up the scsi bus if they were on the same chain as HD. I have had up to 3 compaq controllers(duplexing) in one server with no problems as well as 4 nics. When configuring controllers in eisa config. I always set the builtin scsi irq as 14. Then set the add-in SCSI cards as 10/11 or 12 if not using mouse for something. Don't let the system set them to irq 9 or 15, I have seen this cause problems. I have also had servers share the IRQ on the scsi cards with no problems. ------------------------------ Date: Thu, 18 Dec 1997 09:11:11 +0100 From: Markus Roedel Subject: Re: Problem with SYS volume >>I have an big problem. >>We use an older server ( IBM Model 80) as e-mail server (Pegasus mail >>and Mercury NLM's). Last week the server dismounted the SYS volume >>one time. On friday it was two times en this week it happens several >>times. (So the time is there for a new server) >> >>With startup the server gives the following messages: >> >> loading module PS2SCSI.DSK >> IBM SCSI Adapter Device Driver >> Version 3.11 Februari 12, 1991 >> Copyright 1991 Novell, Inc All rights Reserved. >>Mounting volume SYS >>Initializing Transaction Tracking System >>scanning TTS Backout file. >>12/17/97 11:14am: 1.1.45 Error writing tot the directory on /SYS. >>12/17/97 11:14am: 1.1.45 Error writing tot the directory on /SYS. >>12/17/97 11:14am: 1.1.47 Error Reading both copies of the directory >>on /SYS >>12/17/97 11:14am: Device #0 (26060) SEAGATE ST52151N deactivated >>due to drive failure. I had exactly the same problem with a SCSI-cable that was not built to the SCSI2 specs, eg. the distance between the connectors was less than 15 cm. It worked fine until I added a mirrored drive to the sys volume. The increased traffic on the SCSI bus caused the error mentionend above, but only at times of high traffic, during remirror or when doing an ncopy. Turning off read after write verify also let the problem disappear. Maybe there is still the chance that not the disk, but only the cabling is broken...any recent changes made to the server? ------------------------------ Date: Mon, 12 Jan 1998 22:04:26 -0600 From: Mike Engel Subject: Re: Seagate and Adaptec incompatibility >*** Problem: >I get Page Fault Processor Exceptions (Interrupt Service Routine and other >running processes), Invalid Opcode Processor Exception, and a couple of other >abends during multiple installs or while sitting idle running NW. > >*** Troubleshooting process: >I have done the following: >- checked all IRQ & mem addresses for conflicts >- exchanged RAM >- exchanged SCSI controllers (same make and model) >- installed without NIC to eliminate it as the culprit. >- applied all recent and relevant patches after a recovered install >- installed another identical server (except it has one Conner CFP2107S HD >instead of a Seagate) and it has been running just fine for a couple of months. >- termination on controller and drives have been checked. > >I have a friend who has had problems with Seagate and Adaptec compatibility. He >fixed his problem by exchanging his drives for a different model. Has anyone >else experienced this? I had very similar problems with almost an exact server config, (except Quantum Fireball Drives) After pulling most of my hair out it was solved by slowing the DMA speed on the SCSI controller. If I remember correctly, the card defaults to 40mb/sec. I slowed DMA speed to 20mb/sec and problems disappeared. ------------------------------ Date: Thu, 15 Jan 1998 16:07:36 +0000 From: Richard Letts Subject: Re: Cables >Finally, with SCSI problems, it could always be the cable. Buy a >Granite Digital or Amphenol cable, even if you have internal drives. Our backup server was down for almost a week with a bad ultra fast wide SCSI cable to the external RAID array. Good cables are worth the price. ------------------------------