A pc came for repair yesterday with a most unusual fault.. It started to boot xp and then BSOD with an UNMOUNTABLE_BOOT_VOLUME error..
I think "strange" and do the usual.. boot xp install disk and enter the recovery console... a quick "fixmbr" and "fixboot" and that usually does it...
Not this time.
One quick boot of slax killbill and all the users data and in fact everything else is happily backing up onto a 60 gig usb drive I keep for the use of...
Then further investigating.. It's weird.. because everything is backed up I decide to have a further check of the drive and filestructure.. Linux has no problem accessing the files, and everything looks fine.. running the grub disk shows the MBR is all present and correct for a single partition XP machine.
So try running gparted.. the partition isn't formatted.. Uhhhh??
Format partition NTFS.. copy files back on from backup drive.. set partition bootable, run fixmbr and normality is resumed.. So what the hell just happened???
(I'm expecting a bounce on this machine.. Not happy that the real cause has been found and fixed yet)
Once again, a live CD saves the day. Strange that the partition reported unformatted, but it still mounted under Linux... Was it maybe a case of the filesystem type code got corrupted, but "mount -t autofs" worked it out? I know that when installing Windows, if a partition is given a code 82 or 83 the installer doesn't see it at all, whether it is formatted or not, so that might have just made it go "invisible"?
Sounds about right. I mentioned it to a colleague today and he has seen it before.. Not being a linux user he was forced to just reformat and reinstall, so the trick seems like a good way to rescue data in the event (he thinks anyway) that the file table has become corrupted.
It looks like slax doesn't use the existing file table and type data.. rather it looks at the device as unknown mass storage and makes an "educated guess" as to the filesystem depending on what it finds.
Maybe I should drop a link to all this in the slax forums and see what they think.
The partition wasn't invisible, it was just unformatted.. with the usual primary large partition and the typical xp "free space" at the end so I agree.. It does sound like the type information had been lost somewhere.. Is this what DSL does sometimes if run on an xp machine?? That's worth thinking about too.
It's still odd how I was able to get all the files from the drive (ntfs) but nothing else reported it as formatted. Every partitioner reported unknown, gparted, qparted, grub megadisk, cfdisk on my stripped dsl disk and the tools on both hirons and UBCD .. somehow it had lost the filesystem type. The killbill just mounted it read only, no input from me at all.. I really should have tried puppy with Mutt just to see, but time was late.
I still have a funny feeling I will see this machine again.. whatever caused it last time will no doubt happen again soon. In the meantime I think some experiments with a small working xp ntfs drive are in order to try and duplicate this really weird fault.. .. just don't quite know how to go about it apart from possibly "unformat" from a command prompt without windows running...
The owner doesn't know enough to do that.. could there be some virus which does this?
They had just installed AVG free antivirus and run it apparently.. everything was good, no nasties found and they rebooted next time to the error.... hmmmm..
I tend to find that strange hard drive errors are more a symptom of cheap power supplies. In my experience at least. The vast majority of everything tends to trace back to cheap dodgy ram or a cheap dodgy power supply. Man, everyone I associate with is a bum... Anyway, a little power ripple here and there can result in a drive head doing some strange writes to strange places...
I've personally had a 220W PSU in a small form factor Shuttle case never give me issues for years, despite having two hard drives, DVD, top of range video card and CPU, and I've seen 500W PSUs powering just a hard drive, an all in one mobo and a DVD drive flake out after just a couple of months. The build quality and heat dissipation of the Shuttle PSU was just top notch. Cheap PSUs and ram just aren't worth the risk, nothing can kill system stability more than having those two parts cheap and nasty.
Also worth thinking about.. I didn't open the case as I wasn't at home and everything seemed to be working. When it was running everything seemed stable.. cpu ticking over nice at around 3% idling, and no large spikes.. ram usage in the normal 180-220 range. I did set the smart on for the only hdd.. a seagate as well.. I know maxtor drives can pull funny turns before they die, but seagate in my experience start to show lots of bad sectors first, and will go on for literally years with some bad sectors.
My OS drive on this has over 40 bad sectors and is fine. Just throws a few failed crc check errors on boot. (old seagate rubber jacket xbox drive)
The ram checked out perfect.. I did test that with one of the ram test tools on UBCD
Will scandisk the suspect at the next opportunity.. if it comes back that is.. I have a habit of fixing them well.
It's not a new machine.. P4 1.6 with 512 ram.. seems very well put together, and came from a business so I think the hardware is top notch..
Both cd drives are perfect, and no fans rattling.. case doesn't ever seem to have been opened since it was built.
Come to think of it.. the wall socket did seem a little dodgy.. the modem was going on and off occasionally.
Will carry out further investigations next time it goes mad.
[edit].. still running well.. no sign of the problem coming back [/edit]
