Major SCO Unix issue, please help.

[slodave]

Hi!

My background is mainly in the various Linux distros, but I have picked up a little knowledge over time in SCO unix. Enough that I help out a Hydraulic co., that makes a lot of parts for the military. They are using an ancient 486 server with SCSI drives and SCO. Early morning Friday, a keyboard extension cable shorted - good thing it did not catch anything on fire - and killed the MB and possibly the HDD as well. The main SCSI drive is starting to click and sometimes is not recognized. I have used a device once before, but do not have access to it any more and was able to clone the drive once before when it failed.

Now I am trying to use Norton Ghost, but it is not seeing any SCSI drive, even though I am loading SCSI drivers. That's one problem.

The other problem, is that it seems that SCO does not like having disk geometry changed. How can I duplicate the existing HDD to a new (old) SCSI drive, that does not have the same geometry?

Any ideas would be most helpful. I need to call the company in the A.M. and give them an update. I would really like to get this done this evening though. They will be switching to a Windows based software system in the next couple of months, so I just need the SCO box to run a little longer.

Dave

[Scott R]

Originally Posted by slodave

Hi!

My background is mainly in the various Linux distros, but I have picked up a little knowledge over time in SCO unix. Enough that I help out a Hydraulic co., that makes a lot of parts for the military. They are using an ancient 486 server with SCSI drives and SCO. Early morning Friday, a keyboard extension cable shorted - good thing it did not catch anything on fire - and killed the MB and possibly the HDD as well. The main SCSI drive is starting to click and sometimes is not recognized. I have used a device once before, but do not have access to it any more and was able to clone the drive once before when it failed.

Now I am trying to use Norton Ghost, but it is not seeing any SCSI drive, even though I am loading SCSI drivers. That's one problem.

The other problem, is that it seems that SCO does not like having disk geometry changed. How can I duplicate the existing HDD to a new (old) SCSI drive, that does not have the same geometry?

Any ideas would be most helpful. I need to call the company in the A.M. and give them an update. I would really like to get this done this evening though. They will be switching to a Windows based software system in the next couple of months, so I just need the SCO box to run a little longer.

Dave

Did you try Acronis? We can't image SCO with Ghost at all, which is our primary image platform (or VMware for that matter) So on our unix/linux stuff we use Acronis. Of course all of this academic if you can't even recognize the controller. I'm guessing it's an older Adaptec 2900 series possibly?

[slodave]

I have not tried Acronis yet, but will. You are right on the SCSI cards. All the Adaptec cards I have tried are 2900 series. I do have one SIIG AP-10 Fast SCSI2, but that did not get recognized either. I wish I had the old cloning machine. It was slow, but would duplicate almost anything.

Dave

[stomachmonkey]

Dave,

Clicking SCSI drive coupled with no recognition sounds like a termination issue. Check the drive, card and cable.

Any of the three are capable of providing termination, if more than one of them is you'll have issues.

A terminating cable is easy to spot. My experience with cards is they do not provide termination by default, at least not on the external bus.

Start with the drive.

If I had to guess I'd say the jumpers are not set to terminate.

Any other SCSI devices in the chain that could cause an ID conflict?

The 2900 series is pretty vanilla, I'd stick with it.

WHere is the drive? External case or are you putting internal on the PC?

[slodave]

The drive seemed to click a few times while I was at the business and at first, was not recognized, then it was. The owner also said that the machine seemed to faster on some days. I have everything hooked up to a temporary (faster) machine, but now I can't get the MB to boot my cd first. This is minor though.

As far as termination, The server has been running without major problems in its current config for years. I have only had to save the SCSI drive once, but strangely, I see no terminator on the cable at all. The devices normally connected are 1) SCSI drive (OS), 2) removable backup SCSI drive, 3) SCSI CDROM I have not looked, but can this auto terminate?

Acronis did recognize two other SCSI drives I was playing with, but neither has an OS to copy over and one of the drives is the spare backup HDD they keep just in case the main OS drive and it is bad now. Yikes, they should have migrated many months ago.

Wayne, I am all too familiar with killing HDD's myself, but they do have a backup of the important data, they would lose only one days work.

After all this is done and they are on the new system, I may see if I can keep their old server and frame it. It's a real piece of history , too bad it smells like hydraulic fluid. <- And now my house is starting to smell too.

Thanks for the ideas so far. Looks like it is time to walk to the store and get some beer. This may be a long night. I also have another drive I need to recover as well.

Dave

[Scott R]

Originally Posted by Wayne at Pelican Parts

At this point, if the data is very important, I would recommend next-day shipping it to a recovery company. You might inadvertently mess something up while trying to get it back (been there, done that).

-Wayne

This is great advice, for anyone, and I know Dave already knows this. When you have critical customer data on a questionable drive, send it out. places like "Ontrack" http://www.ontrackdatarecovery.com/ Can take that drive apart in a clean room and rescue it. Of course this is little help to SCO operating system looking for a boot loader, however still a good thing to do.

[slodave]

I have been doing recovery on hard drives for a long time and know when it's time to send it off. This is also why I am trying to get everything setup with other drives first, so I don't put additional strain on the potentially bad drive.

I learned a personal lesson a number of years ago. I wanted to copy a bunch of vacation pics from my laptop to one of my Linux servers for a backup, but I accidentally moved them and one day, the server crashed. The platters were actually scratched from the heads. Total loss.

The pros have a lot of tricks before they go to the clean room. One is to switch circuit boards from a good identical drive

Dave

[stomachmonkey]

Are you working with the original server or is everything hooked up to a different box?

SCSI is great but there is a bit of vodoo to it.

If everything has been moved you need to a) check for conflicts b) check for termination. Termination is easy, kind of. Basic rule is the last device in the chain gets terminated. Device ID's are set in order of the chain, 1st device having the lowest number, usually.

It is possible to have some devices in a chain show up and not others even though the device works.

Terminated cables are easy to spot, they will have a big block on the end. Won't look like a normal ribbon cable.

Checking termination on drives is easy, look at or around the pinouts, the area on an IDE where you would set master/slave. The drive should have a schematic telling you what pins are what, sometimes it's printed right on the board. Termination is usually labled "term"

SCSI Id's range from 0-7. The ID is acheived by jumping 3 pinouts. No jumpers = 0, jumper first set only = 1, jumper second set only=2, jumper third set only= 4. Jumping combinations gets you 3,5,6,7. jumper 1st+2nd =3, 1st+3rd=5, 2nd+3rd=6, all 3=7.

CD drives like to be either 1st or last, not in between. Boot drives should be set to 0.

Now if everything is still in the original server and the boot drive does not show I'd say it's fuched.

[Scott R]

Originally Posted by slodave

I have been doing recovery on hard drives for a long time and know when it's time to send it off. This is also why I am trying to get everything setup with other drives first, so I don't put additional strain on the potentially bad drive.

I learned a personal lesson a number of years ago. I wanted to copy a bunch of vacation pics from my laptop to one of my Linux servers for a backup, but I accidentally moved them and one day, the server crashed. The platters were actually scratched from the heads. Total loss.

The pros have a lot of tricks before they go to the clean room. One is to switch circuit boards from a good identical drive

Dave

Yea, always good to have spare parts like that on hand. ;

[slodave]

The good... Acronis sees two drives and all should be going well...

The bad... The spare SCSI drive the company has, may be bad as well and the only other SCSI drive I have is too small, well, I have one other drive, but no adapters for it...

The ugly... From what I can tell, there is no practical way to image the drive to PATA or SATA, without causing SCO to freak on boot. If there is a way to do this, I can give them a temp machine and ditch SCSI. I found something on Google that seemed to say that there is some commands you can issue on the existing drive that'll allow this to be possible, but I may be wrong. Any one know for sure?

Off to investigate one last SCSI drive I have.

Dave

I just got a call from someone and from what they said, it sounds as if the HDD failed, that makes three in a row.

[slodave]

Originally Posted by stomachmonkey

Are you working with the original server or is everything hooked up to a different box?

SCSI is great but there is a bit of vodoo to it.

If everything has been moved you need to a) check for conflicts b) check for termination. Termination is easy, kind of. Basic rule is the last device in the chain gets terminated. Device ID's are set in order of the chain, 1st device having the lowest number, usually.

It is possible to have some devices in a chain show up and not others even though the device works.

Terminated cables are easy to spot, they will have a big block on the end. Won't look like a normal ribbon cable.

Checking termination on drives is easy, look at or around the pinouts, the area on an IDE where you would set master/slave. The drive should have a schematic telling you what pins are what, sometimes it's printed right on the board. Termination is usually labled "term"

SCSI Id's range from 0-7. The ID is acheived by jumping 3 pinouts. No jumpers = 0, jumper first set only = 1, jumper second set only=2, jumper third set only= 4. Jumping combinations gets you 3,5,6,7. jumper 1st+2nd =3, 1st+3rd=5, 2nd+3rd=6, all 3=7.

CD drives like to be either 1st or last, not in between. Boot drives should be set to 0.

Now if everything is still in the original server and the boot drive does not show I'd say it's fuched.

I am using a more modern computer. I have no interest in working with a 486 server with 64 megs RAM . The computer I am using only has a IDE cdrom to boot my wonder cd and a SCSI card for the two drives. Same cable from server, but terminated after second drive with my terminator. The cable does not have an external one and nothing built in.

Dave

[Scott R]

Originally Posted by slodave

The good... Acronis sees two drives and all should be going well...

The bad... The spare SCSI drive the company has, may be bad as well and the only other SCSI drive I have is too small, well, I have one other drive, but no adapters for it...

The ugly... From what I can tell, there is no practical way to image the drive to PATA or SATA, without causing SCO to freak on boot. If there is a way to do this, I can give them a temp machine and ditch SCSI. I found something on Google that seemed to say that there is some commands you can issue on the existing drive that'll allow this to be possible, but I may be wrong. Any one know for sure?

Off to investigate one last SCSI drive I have.

Dave



I just got a call from someone and from what they said, it sounds as if the HDD failed, that makes three in a row.

Dave, you should be able to do this, however the solution is painful. If you have a ESX server you can P2V (VMware Converter) the Acronis image to a vmdk file. After that it can be restored in the virtual environment and moved to whichever hard drive platform you want to use. NOt the best solution I know, but we use it for DR from time to time.

[pmajka]

I used to deal with a Mom and Pop shop....i486, 800Mb scsi drive...I mentioned raid, tape backup solution and keeping the Long Hair'd cat out of the office.

well they lost all their data...root cause, power supply smoked due to fan failure and hard disk would not spin again. Cause: Long Hair'd cat hair clogged fan, PS and MB!

lesson learned, if they wont dump a measly Grand in computer HW...walk away.

Good luck with recovery...I know it can be a nightmare.

[slodave]

Here's where it stands. The current HDD is 9 gigs. The spare that I took is only 3 gigs and is bad. I have two SCSI drives that were supposed to be taken apart for their staters, but never got a round to it. One is 2 gigs and the other requires a cable and card that I don't have. The current drive is not making funny clicking noises anymore. I wonder if termination has been a problem all along, but they have been getting lucky.

I no longer have the option to get another SCSI drive tonight, and really don't want to have them spend money on an over sized SCSI drive that'll be taken out of service in about two months anyway.

Scott, have any leads on the hard way. I don't have access to a ESX server.

[stomachmonkey]

So are these external drives? Not internal?

Some drives have a term toggle on them.

Is your terminator active or passive?

All academic as I think the drive is toast. If it was clicking when the server was functioning then, well, you know.

[slodave]

Originally Posted by pmajka

I used to deal with a Mom and Pop shop....i486, 800Mb scsi drive...I mentioned raid, tape backup solution and keeping the Long Hair'd cat out of the office.

well they lost all their data...root cause, power supply smoked due to fan failure and hard disk would not spin again. Cause: Long Hair'd cat hair clogged fan, PS and MB!

lesson learned, if they wont dump a measly Grand in computer HW...walk away.

Good luck with recovery...I know it can be a nightmare.

Even thought they are a small shop, they have a number of government contracts and have the money to spend. It's just that the new software they are using now, can do everything and the new software from the same company (to the tune of $100,000 including the new server) leaves out a key component - payroll. There have been a couple of other companies out there holding on as well, but the SW company has finally terminated support. They do a good job of backing up and will only lose one day, it's just that the backups do no good without a functioning SCO OS.

I like fighting with Unix. Not many people left that know anything or want anything to do with it and it pays well

Anyway, I'm off to the Quick-E-Mart.

Dave

[stomachmonkey]

Originally Posted by slodave

Here's where it stands. The current HDD is 9 gigs. The spare that I took is only 3 gigs and is bad. I have two SCSI drives that were supposed to be taken apart for their staters, but never got a round to it. One is 2 gigs and the other requires a cable and card that I don't have. The current drive is not making funny clicking noises anymore. I wonder if termination has been a problem all along, but they have been getting lucky.

I no longer have the option to get another SCSI drive tonight, and really don't want to have them spend money on an over sized SCSI drive that'll be taken out of service in about two months anyway.

Scott, have any leads on the hard way. I don't have access to a ESX server.

Dave,

I'm a pack rat with "that stuff" and keep it forever but am in the middle of moving.

I just tossed a whole bunch of SCSI cards and known good 9gb Barracuda SCSI drives.

Now I wish I had not as it seems you could have made good use of it.

Sorry.

[slodave]

Originally Posted by stomachmonkey

So are these external drives? Not internal?

Some drives have a term toggle on them.

Is your terminator active or passive?

All academic as I think the drive is toast. If it was clicking when the server was functioning then, well, you know.

There is one removable enclosure, but I think it is second on the chain, but I would have to look. I am pretty sure that the drive made a few clicks, but only around spin-up. I know that most SCSI drives tend to click at least once when the heads are released. Maybe what I heard was both drives releasing the heads?

I'd just rather be safe and clone the drive. Looks like I hold off and try to find an old similar drive around town or I figure out how to clone it to PATA/SATA...

Dave

[pmajka]

I know this will sound funny, but i am not joking here..

Buddy of mine had a system at office, that ran on a 486 (companies, not his).... the payroll software wouldnt run on anything but Windows 3.1. Had no desire to move to modern system.

HW was on last leg. He took P3 running W2k, with Raided boot disk, Ran an I486 emulator program and moved the Win3.1 and Payroll on to it. modern HW running the OLD os...something like 500X faster.....but safe.

time for bed.

[Gogar]

Dave, I've got a closet full of Ultra-2, Ultra, and Reg SCSI drives and enclosures in my closet, with adapters, terminators, cables, you name it. Come get 'em. I think i've got 6 36GB drives ready to go.

Oh, wait.