Miscellaneous Operating Systems/Hardware

Anyone ever try this (successfully) ?

I lost a disk a while back. It was backed up but not totally current so it's not a total disaster but there was some stuff I wouldn't mind saving if possible.

Found an exact duplicate, got it for ten bucks. Tried the circuit-board changeout method, with the new (good) board in the bad disk the disk spins up but gets the click of death and isn't seen by hinv or fx. Also tried the bad board in the new disk (tested the disk first, it's fine) and no go. Swapped the boards back and the replacement works again. So the circuit board is definitely bad but so is something internal.

Pulled the cover off the bad disk and took a look - what the heck, it's bad anyhow - and nothing looks broken. No scratches or marks of any kind in the platters. (2 platters.) So it seems likely that something croaked in the head driving motor and then took out the circuit board along with it. Luckily, only ran the new one a very short time.

It seems like a person could swap out the platters and if careful, it should last long enough to get the data off ... for ten bucks I don't mind trying but if this is impossible, it's not worth wasting a lot of time over.

Anyone here ever succeed at this ?

No one's ever taken a disk apart ? I'm ashamed of youse guys ...

Mighty Casey has struck out but did learn some stuff :

In this case, I was barking up the wrong tree from the very beginning : the disk that failed was a high-quality enterprise-grade 15k Fujitsu scsi disk that only had a year or two of service. It dates from 2006 but hadn't been used much. it was working fine when I shut the computer off, then never showed up the next time it was booted.

Smart first move would be to run fx against it. Unfortunately, as a boot disk and the only 80-pin sca disk I had, this was impossible. I got totally off on the wrong tangent trying 68 pin disks, cables, dismantling the board that the disks plug in to, etc etc. I kept assuming that it was just a corrupt filesystem and put all my efforts into the wrong places until I finally got a second adapter card installed and tried an fx against that disk separate from the boot controller. Oops.

If you have a second way to access a disk, that's a big help in diagnosis. Another SGI or a Loonix box that can read xfs would do. I put too much trust in the fact that this disk had been fine when shut down and made no bad noises. fx is your friend.

The basics are readily available by search but here's what I noticed :

If you are lucky, this isn't anything to be afraid of. All those "oh noes ! you will instantly destroy your disk if you open it !" pronouncements are bullshit. Yes, it for sure won't last five years after opening. But it's already dead. Don't be a baby, all the parts are made in non-clean-room environments. They aren't vampires, exposure to the sun won't melt them into a blob on the floor.

If you are lucky, a logic board swap will fix the problem. It's easy. Yay.

If that doesn't work, I'd consider pulling the cover off and powering the disk up in the open. I didn't do this and ended up chasing the wrong rabbit because of it. I thought the disk was spinning up because the case got warm. In fact, it wasn't. Swapping out the heads when the problem is the motor is not going to fix anything.

With the cover off, you can see that the disks don't spin up until they are addressed. I always thought they spun up at power-on but no. You can watch the bootup terminal display; when the process gets to each adapter, then the disks are powered up. Not sure if this is true for the boot disk but observed it on a secondary adapter.

Next step would be to swap out the heads. This is not so difficult. I used strips of paper to slide them off the platters without crashing. That was pretty tacky but it worked. Six hands like those Indian godesses would be a help. The twenty-six boobs, not so sure about.

The magnet is very strong, when you first remove it it will surprise you and try to leap at the disk, bringing about instant destruction. Be prepared.

Replacing the heads is actually pretty simple but decent tools would be a help. Toothpicks, pieces of paper, melted soda straws do not count

All that work and the thing still didn't show up, crap. So I took the cover off and viola, the disk wasn't spinning. Oh phooey. This is why I mention taking the cover off early in your crusade might be smart. I think swapping the heads an extra time was not an aid to successful surgery in my case.

If you have a single platter, swapping it would be easy. If you have two or more, you're screwed. They need to be timed but there is no physical feature doing this. I looked online, there are platter-removing tools but the ones I saw were very expensive pieces of crap. $ 300 for a tin can with two screws ? those guys must be crazy. If the data were important you could make a decent tool - maybe even with a 3d printer Otherwise draw something up and have the local shop mill it out for you out of delrin. My unbacked-up data was just some customizations I had done to the desktop and some emails that are stored offline anyhow. Of course I can't get to offline because the gfw has it blocked but that's a different problem. In this case, I just knocked out a crappy little thing to try to hold the platters in time to each other. Didn't work very well but if you have fairly immportant data, it should be easy to accomplish.

Swapping the platters is even easier than swapping the heads (except you have to get the heads out of the way anyhow.)

In conclusion ... this didn't work for me but that was because I had crappy tools, made some wrong conclusions, and wasn't that concerned about the data. If the data was important but not important enough to spend a ton of money on recovery, chances are good you could do it. And the clean room thing is hogwash. Sure, for a disk to last five years you'd need that. But to run foir twenty minutes while you get most of your files, don't be silly. A can of air and a reasonably clean environment wll work fine. Cleanliness is not the biggest problem. In fact, running the disk open will tell you a lot about what's broken, before you waste time and effort and potentially breaking something that isn't the problem. Dealing with very tiny, very fragile parts is the biggest trouble. And keeping the platters timed to each other, thats the biggest problem.

btw, a data recovery service .... this isn't as difficult as they make it out to be. If it's a physically bad disk that isn't crashed or damaged, and it's not a file-system problem, getting the data off should be doable for a careful amateur. There's gold in them thar hills.

Is it a job for Herr Doktor Rotwang?
I've taken old disks apart but not with the intention of fixing them. From what I gather the torque applied to the case screws is also somewhat critical? But a reasonable torque driver is pocket change, it shouldn't take aerospace tools.
In the days of PDP11s and such things it was much more common to repair hard drives, since they only lasted a few thousand hours between failures. Obviously no clean rooms.

robespierre wrote: In the days of PDP11s and such things it was much more common to repair hard drives, since they only lasted a few thousand hours between failures. Obviously no clean rooms.

I've fixed a many RA81 and RA82's, gotta love those 150 pound winchester drives

hamei wrote: With the cover off, you can see that the disks don't spin up until they are addressed. I always thought they spun up at power-on but no. You can watch the bootup terminal display; when the process gets to each adapter, then the disks are powered up. Not sure if this is true for the boot disk but observed it on a secondary adapter.

For what it's worth (probably not much!) that's usually controlled via jumpers on the drive itself. If you have a large number of drives in a single enclosure the in-rush current from them all spinning up at once could be too much for the power supply to handle (even though it would be fine with them all running in a steady state). So the disk manufacturers provide a way to avoid the problem.

The manual for your exact drive could tell you more, but my Seagates have two jumpers to control spin-up. Normally no jumper is installed and the drives all spin up immediately at power on. Or you can put a jumper in one position and the drive will wait for the host to send the SCSI "Start Unit" command. (Which is likely the behaviour you saw.) Or you can put a jumper in another position (maybe a host doesn't send the "Start Unit" command) and the drive will power itself up after a delay that's calculated by taking the drive's SCSI ID number and multiplying it by 12 seconds (so drive 0 after 0s, drive 1 after 12s, and so on). Either way, the delayed spin up avoids problems caused by needing too much current at power on.

I think it's kind of clever and shows that some thought went into designing these things for a variety of situations.

robespierre wrote: Is it a job for Herr Doktor Rotwang?

It's not very well known but the good Doktor did make a series of technical films describing his methods . I did watch several times before turning the first screw ...

I've taken old disks apart but not with the intention of fixing them.

Didn't give myself better than 50-50 odds here but what the heck, can't hurt to try ...

From what I gather the torque applied to the case screws is also somewhat critical?

At least on this Fujitsu, the body is a one-piece casting with a screwed-on sheetmetal cover. I can't think the torque applied to those is especially critical.

In the days of PDP11s and such things it was much more common to repair hard drives, since they only lasted a few thousand hours between failures. Obviously no clean rooms.

Ja, I did a pulley change on a Shugart once to convert from 60hz to 50hz. And we had the disk refaced and plated ... even these new disks are not that touchy, I think. After I butchered it about four times the victim did show up on startup but since I got the platters out of sync the filesystem was trash. If it were more important it seems like a careful amatoor would have a fighting chance.

jpstewart wrote: For what it's worth (probably not much!) that's usually controlled via jumpers on the drive itself.

I've seen that on 68-pinners but checked all the 80 pin disks I have here, none of them have spinup or address jumpers. After playing with this thing tho, I am more impressed with SATA drives.

I think it's kind of clever and shows that some thought went into designing these things for a variety of situations.

I'm not sure what I think now ... the drives are rpetty nice but it seems like grossly overpriced - notice how the prices for $300 "enterprise" drives have crashed through the floor ? And I have a sneaking suspicion that there's really not much difference between a garden-variety disk drive and the high-dollar spread. You pay extra for the warranty, statistics takes care of the rest. Have you noticed how hard it is to come by real comparative statistics on drive life ?

I wasn't all that excited about SSD's - a hard disk is fast enough for me - but after having five of these things fail without warning in the past six months, I am becoming more fond of the church of solid state.

Anyway and in conclusion, whether you can retrieve the data off a failed disk is going to depend on exactly what failed but I don't think it's nearly as impossible as many people claim. The most difficult thing is to keep the platters aligned. I did a little searching and saw the abortions people were calling "professional" tools - mary joseph and the baby jesus, what a bunch of grossly overpriced crap. So I went ahead and over-risked it. But thinking later, something like this (for two platters) :


would be easy to build and should do the job. The dimensions would have to be modified to fit the particular platter setup, and some experimenting with materials would be in order (UHMW might be good. Delrin too stiff and scratchy ? maybe some sort of Nylon ?) - I put a screw crosswise in the base to stiffen the anvil, don't know if that would work or not but can try. Anyway, I'd bet this would work better than the crap they sell and would cost about $15 to make. Might even be able to 3d print one ... don't forget to break the edges. Here's an stl to play with, if it can do anyone any good ...