I loved my Sun 6 bay 18 GB SCSI array on my Sun box, but I can't get it to work on Windows XP?
November 2, 2005 10:38 PM Subscribe
Yeah so I went the complete cheapo route and bought an Initio KM-910UW SCSI card from TigerDirect.com and I'm pulling the rest of my hair out. I can successfully see all 4 fujitsu and 2 seagate 18 GB drives in their Ctrl-I (like Adaptec) BIOS. I've got zero conflicts in Device Manager and my Events are so clean minus what's mentioned below.
No matter what combinations I try, e.g. disks/slots/drivers/try or buy software from download.com. I'm unable to get more than one drive to format and work correctly. I've been able to format in any slot w/only one drive but as soon as I add a second or even a third or all six, the second drive will refuse to format and/or the system will probe to no end thus hanging the boot process or after boot get a BSOD or chkdsk after a reboot on the 1st drive...
The main error in event viewer I get is that the card is not replying during the timeout limitation - "The device, \Device\Scsi\ini910u1, did not respond within the timeout period." Unfortunately my main computer BIOS (or the SCSI based BIOS) has no threshold for me to increase as far as timeout values go.
My first hope is that b/c I'm such a Windows hater is that I'm missing something there. I'm a Sun SysAdmin and ex Oracle programmer/pseudo DBA.
I'm desperately hoping that someone, somewhere has been able to get a Sun 6-drive Multipack to work w/a Windows XP IDE-based workstation.
I'm properly terminated, I've tested so many combinations of drives, cables and slots it's disgusting.
Thanks in advance for _any_ ideas, including a new Adaptec card that will hopefully do the trick.
Windows always defeats my Unix logic...
I can't figure out if there is a correct answer here or not.
posted by AllesKlar at 11:43 PM on November 2, 2005
posted by AllesKlar at 11:43 PM on November 2, 2005
Looks to be a few things on the m$oft site you might not have tired
posted by AllesKlar at 11:46 PM on November 2, 2005
posted by AllesKlar at 11:46 PM on November 2, 2005
Wow; tried / tired; it's all very telling. So are you formatting the drives in windows or through the scsi card bios? When you say the boot process, you mean shortly after POST or whilst loading XP?
posted by AllesKlar at 11:51 PM on November 2, 2005
posted by AllesKlar at 11:51 PM on November 2, 2005
Best answer: I have 3 Maxtor U320 drives running on an Adaptec 29160N card on a Windows XP SP2 box, and several Adaptec 2940UW cards working on UW SCSI chains on Debian and FreeBSD boxen. I looked at the Initio brochure on the Verbatim site, but there isn't much "technical" information there. Still, it sure sounds like you're having a termination problem, leading to bus arbitration failure. The fact that you can "see" the devices in the Initio SCSI BIOS doesn't mean much, as unless there is a device numbering conflict, the BIOS enumeration routine for discovering devices may not call for any of them to actually arbitrate the bus (at least, it doesn't on Adaptec or LSI cards), as they must when actually going through boot and operation. This theory is particularly likely if any of your devices are different SCSI generations, i.e. mix of SE, Ultra, and U/W, or are running in an external enclosure. Basically, you have to cable and terminate for the highest standard on the chain, to allow the fastest and most intelligent device to successfully arbitrate it's way "down" the SCSI protocol chain, to have them all work happily with the other devices. This, of course, includes the adapter card itself.
Earlier generations of multi-protocol Adaptec cards, like the 2940U2W, were notorious for having self-termination problems, which were partly solved by using U/W active termination at the cable ends, and partly by firmware fixes that helped the active termination on the cards themselves work properly, particularly when the cards were used in the middle of a SCSI chain, or with SCSI devices of different generations on external and internal connectors. In my experience (echoed as gently as possible by every SCSI vendor's tech support folks..:-\), 99.678435% of all SCSI problems are resolved by fixing cabling and termination issues.
So, it sounds like it's time to get medieval on your SCSI chain. Start with a fresh cup of coffee, a clear head, and pretend you've never seen this hardware before. Begin by removing every drive on the chain, and manually verify each configuration jumper, and make written notes as you do. If any drive is jumpered to supply termination power, change it to not do so. Verify that every drive does, in fact, have a different SCSI ID number, and that none of them are 0 or 7, as your card could be reserved to one of these. Check the cables and connections again, particularly if you have single end, Ultra, and UltraWide devices mixed on the chain, and are using 50-68 pin adapters, or have any SCA drives. Examine the backplane of your enclosure carefully, and verify physically all cabling is per recommendations for the SCSI standard for the "highest" generation of SCSI device on the chain, including the adapter card. Since your Initio card appears to be UW, this means your whole chain needs to be UW (68 pin).
Before connecting any SCSI devices, verify that your card has the latest vendor firmware, and if it doesn't, update it. Then go into SCSI BIOS, and verify that the termination of the card is set for active termination on UW. Don't trust the auto-termination on the adapter card at this point, since if you are having physical termination problems, the card's auto-termination routines may not be properly sensing what is required, which is why you're carefully setting termination manually. Install only one 68 pin active terminator at the end of your physical chain. Don't try to use any of your drive's internal termination to save a few bucks on termination. At this point, only the adapter card and the active terminator on the end of your chain are supplying SCSI bus power and termination. If you are using and external enclosure, you may have to adapt your cabling/termination strategy as necessary to the hardware you have, but this is the necessary condition you need to achieve. The card is at one end of your chain, there are no devices on the internal connector, all the SCSI devices are on the external connector, there is one and only active termination at the end of the external chain, and the card is supplying active termination at it's end, and is the sole source of bus power. Assuming you've observed the physical limits for UW, and have this all set up correctly, you won't have problems.
Unless, of course, your card is still incorrectly terminated, itself.
I've saved this point until last, only because your card appears to be physically the same 3 connector layout as the infamousAdaptec 2940U2W.
Early versions of the 2940U2W caused all kinds of grief, because people unwittingly assumed that auto-termination would actually work to save them the trouble of figuring out how to setup their SCSI chain, while Adaptec assumed there would always be devices connected on the internal connectors, and only sometimes would the card also support external devices. So, long story short, for these early cards, to get an "external only" SCSI chain to work, you had to treat the card as if it were in the "middle" of the chain (thus disabling all internal termination on the card), plug in an actively terminated dummy chain cable on the internal 68 pin connector, manually set or disable device types for each SCSI target on the card BIOS, and then connect your actively terminated external chain. I think this was later fixed in silicon, and the U2W variant is no longer actively offered by Adaptec, but huge confusion still reigns regarding the mojo required for using this card successfully. Which is why you can pick them up cheap on eBay...:-)
posted by paulsc at 4:49 AM on November 3, 2005
Earlier generations of multi-protocol Adaptec cards, like the 2940U2W, were notorious for having self-termination problems, which were partly solved by using U/W active termination at the cable ends, and partly by firmware fixes that helped the active termination on the cards themselves work properly, particularly when the cards were used in the middle of a SCSI chain, or with SCSI devices of different generations on external and internal connectors. In my experience (echoed as gently as possible by every SCSI vendor's tech support folks..:-\), 99.678435% of all SCSI problems are resolved by fixing cabling and termination issues.
So, it sounds like it's time to get medieval on your SCSI chain. Start with a fresh cup of coffee, a clear head, and pretend you've never seen this hardware before. Begin by removing every drive on the chain, and manually verify each configuration jumper, and make written notes as you do. If any drive is jumpered to supply termination power, change it to not do so. Verify that every drive does, in fact, have a different SCSI ID number, and that none of them are 0 or 7, as your card could be reserved to one of these. Check the cables and connections again, particularly if you have single end, Ultra, and UltraWide devices mixed on the chain, and are using 50-68 pin adapters, or have any SCA drives. Examine the backplane of your enclosure carefully, and verify physically all cabling is per recommendations for the SCSI standard for the "highest" generation of SCSI device on the chain, including the adapter card. Since your Initio card appears to be UW, this means your whole chain needs to be UW (68 pin).
Before connecting any SCSI devices, verify that your card has the latest vendor firmware, and if it doesn't, update it. Then go into SCSI BIOS, and verify that the termination of the card is set for active termination on UW. Don't trust the auto-termination on the adapter card at this point, since if you are having physical termination problems, the card's auto-termination routines may not be properly sensing what is required, which is why you're carefully setting termination manually. Install only one 68 pin active terminator at the end of your physical chain. Don't try to use any of your drive's internal termination to save a few bucks on termination. At this point, only the adapter card and the active terminator on the end of your chain are supplying SCSI bus power and termination. If you are using and external enclosure, you may have to adapt your cabling/termination strategy as necessary to the hardware you have, but this is the necessary condition you need to achieve. The card is at one end of your chain, there are no devices on the internal connector, all the SCSI devices are on the external connector, there is one and only active termination at the end of the external chain, and the card is supplying active termination at it's end, and is the sole source of bus power. Assuming you've observed the physical limits for UW, and have this all set up correctly, you won't have problems.
Unless, of course, your card is still incorrectly terminated, itself.
I've saved this point until last, only because your card appears to be physically the same 3 connector layout as the infamousAdaptec 2940U2W.
Early versions of the 2940U2W caused all kinds of grief, because people unwittingly assumed that auto-termination would actually work to save them the trouble of figuring out how to setup their SCSI chain, while Adaptec assumed there would always be devices connected on the internal connectors, and only sometimes would the card also support external devices. So, long story short, for these early cards, to get an "external only" SCSI chain to work, you had to treat the card as if it were in the "middle" of the chain (thus disabling all internal termination on the card), plug in an actively terminated dummy chain cable on the internal 68 pin connector, manually set or disable device types for each SCSI target on the card BIOS, and then connect your actively terminated external chain. I think this was later fixed in silicon, and the U2W variant is no longer actively offered by Adaptec, but huge confusion still reigns regarding the mojo required for using this card successfully. Which is why you can pick them up cheap on eBay...:-)
posted by paulsc at 4:49 AM on November 3, 2005
My first thought is try all the drives one at a time -- it may be that one drive is negotiating, but not actually answering commands.
Does that SCSI bios has a disk verify? Can it verify all of the disks.
I'd make the obJoke about technical reasons to sacrifice a chicken to your SCSI chain, but you're beyond that.
Hmm -- does that controller have two busses? If so, are two busses showing up in Windows?
posted by eriko at 4:50 AM on November 3, 2005
Does that SCSI bios has a disk verify? Can it verify all of the disks.
I'd make the obJoke about technical reasons to sacrifice a chicken to your SCSI chain, but you're beyond that.
Hmm -- does that controller have two busses? If so, are two busses showing up in Windows?
posted by eriko at 4:50 AM on November 3, 2005
Response by poster: Update: I've since bought a very cheap replacement scsi terminator for my 6 disk multipack array ($6 on ebay - brand new (maybe.)) It just came and I'm having much success minus two drives possibly being bad. Before I did this I researched further and found disk diag software for both my 4 fujitsu and 2 seagate drives. Before buying the terminator I found (and this is strange as I've used this array again recently w/my sun box) that the error associated w/the second drive was definitely termination related, I.E. scsi bus issue.
This issue was 99.678435% termination related and the rest must have been issues w/two bad drives that I'll strive to update upon learning what happened w/them. I'm currently trying to find out the issues w/the two drives, but in the meantime, thanks so much for the elegant post, paulsc. I love having an extra pair of eyes and meticulous mind when encountering issues such as these.
It's so funny b/c when I was a programmer most of the time all it took was either writing down the problem and then going over the notes or describing it in detail to another co-worker and POOF! I ended up explaining the fix to myself. System Administration is a step beyond, requiring another set of eyes and ears and experience to tell you that you were'nt investigating the right areas and/or handing you the fix itself. Thank you. Thank you so much. I'm liking askmefi more every day.
Great blogs too btw... Thanks everyone for your help as well. I followed up on them all and appreciated all your input.
Rob
posted by prodevel at 3:35 PM on November 7, 2005
This issue was 99.678435% termination related and the rest must have been issues w/two bad drives that I'll strive to update upon learning what happened w/them. I'm currently trying to find out the issues w/the two drives, but in the meantime, thanks so much for the elegant post, paulsc. I love having an extra pair of eyes and meticulous mind when encountering issues such as these.
It's so funny b/c when I was a programmer most of the time all it took was either writing down the problem and then going over the notes or describing it in detail to another co-worker and POOF! I ended up explaining the fix to myself. System Administration is a step beyond, requiring another set of eyes and ears and experience to tell you that you were'nt investigating the right areas and/or handing you the fix itself. Thank you. Thank you so much. I'm liking askmefi more every day.
Great blogs too btw... Thanks everyone for your help as well. I followed up on them all and appreciated all your input.
Rob
posted by prodevel at 3:35 PM on November 7, 2005
This thread is closed to new comments.
posted by prodevel at 10:41 PM on November 2, 2005