News:

The anti-spam plugins have stopped being effective. Registration is back to requiring approval. After registering, you must ALSO email me with your username, so that I can manually approve your account.

Main Menu

Hard drive troubles at home

Started by reinder, December 01, 2006, 12:31:30 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

reinder

Here's one for the computer techs around here: I'm having problems with my hard drives; my primary drive has up and died, and my secondary drive is following suit.  Here's a history of what happened.
As I say in the article, I'm not that good with hardware, but I'm prepared to try and fix it myself if I must. Yet before I either take the machine to the shop or rummage in its innards myself, I'd like to have some idea of what to look for. Power supply connections not attached well? Some other problem with connecting it right? Somebody screwed up the Slave/Master settings? Post your hypotheses here, please, and earn my gratitude.
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle

fesworks

QuoteSomebody screwed up the Slave/Master settings?
only if you think someone opened up your computer... that is an external setting.

as far as anything else, I only know how to install a drive and plug it in.

if you look for a new one, be carefell to get the correct connection type... there are a couple out there and they do not connect with the same parts, nor do I think there is an adaptor for it.

www.PSIwebcomic.com
www.TheShifterArchive.com
www.ArdraComic.com
www.WebcomicBeacon.com

reinder

Well they had to open up the PC to insert the drive.... I have no idea how important those settings are anymore, anyway.
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle

dragyn

About all I know is that if you change them wrong, the computer won't boot.  I'm more interested in the programs than the computer's guts.  Sorry I won't be any help.

Xepher

Would've responded sooner here, but I've been offline for the past few days as part of moving. Anyway...

Going on what you've said alone, it sounds like the IDE controller on your motherboard may be failing. The odds of multiple drives failing the same way at the same time like that are pretty slim, much more likely that it's the controller that handles all of them. When you mentioned ReiserFS I thought "Oh Noes!" as the net is full of horror stories about ResierFS eating data for no good reason. However, unless you recently upgraded your kernel or ReiserFS modules in some way, I wouldn't expect it to fail on multiple disks at the same time either. And if your kernel is giving hardware I/O errors... again, not likely the file system. Do you have a way to try those drives in another machine, or at least another controller? I usually use an external USB hard drive enclosure to test drives when I don't have a spare computer. If you can, I'd give that at go, and backup your data ASAP, before trying any further troubleshooting. Most of the things I can think of for you to try can be destructive to data.

reinder

Now that is very useful information - especially if it means the supplier shafted me with a POS motherboard! Thanks.

I'm not afraid of destructive rescue techniques, as long as I can get to my data first. I'll open up my PCs innards in a couple of days and see what I can do.
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle

Xepher

Do what you can to rescue/backup your data first. As I mentioned, try the drives in a different computer, or at least with a different controller (like a USB external enclosure or something.) Once you've done that, I'd suggest you boot of kubuntu (or other live CD) and then try zeroing the drive(s). It WILL destroy any data on the drive though, so don't try it first. "Zeroing" basically means writing zeroes to every bit of the drive, it's often called a "low-level format" and is a good way to diagnose hardware-level I/O problems, as you'll be bypassing all filesystem layer stuff. If you get I/O errors when trying to zero the drive, then it's either a kernel/driver problem, or actual hardware errors. If you're using a Live CD/Kernel/Drivers that you know have worked just fine in the past, then that narrows it down to hardware issues. If you can try the same zeroing process on another computer (or another controller) than you can find out if the issue is with the drive itself, or further up the chain (in the motherboard/controller.)

Another option, rather than manually troubleshooting via linux, is that many hard drive vendors offer diagnostic cds for download, where you can boot them, and they'll run all sorts of tests on the drives. Most of them include a low-level format/zeroing utility/test as well. Check the website of whoever made your hard disk(s) and see what you can find. Just pay attention to the documentation, as some tests on those cds may harm/erase data on the drives, while others are safe (non-destructive) tests.

reinder

I've finally found the time and the energy to try out part of Xepher's suggestions. Today, I bought a IDE-to-USB drive housing thingy, figured out how to remove one of the hard drives from the broken machine, figured out which part of the seemingly simple instructions for assembling the IDE-to-USB housing thingy were wrong and which were impossible to carry out due to shoddy workmanship on the casing, brought the whole shebang to the studio, loaded Kubuntu there and got the drive to work. Yay me! The bad news is that I ended up having some of the same problems that I had with both drives at home: some of the files I knew were there when the computer broke weren't to be found on the  drive, and some of the others seemed to exist, but when I copied them, Konqueror refused to do so. With these, all  I get from copying the files is 0-byte files on the target drive, even though Konqueror lists these files as available and having a believable file size on the source drive. Weeeeird. Opening these files in, say, OpenOffice (I am mainly concerned at this time with my financial accounts as doing my bookkeeping is enough of a pain without having to reconstruct my old invoices. Anything else including music files and even my backed up art images is of secondary importance - that stuff exists on CDs or can be remade easily), I get an error saying the file is corrupt, can't be repaired, sorry, can't open it at all. Strangely, some PDF files I made of several of my outgoing invoices did still work, and these have now been copied on my portable drive, leaving me with only three invoices to reconstruct or otherwise recover. So it's not all bad.

I'm going to do the same with the other hard drive. That one's harder to reach inside the PC's case, but it may have more functional data left on it, including those missing bills and an old copy of my email folder.
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle

Xepher

Yeah, sometimes it's still hard to tell if it was the drive or the controller that died. If the controller (or some other non-drive bit of the system) failed, it could start actually corrupting the filesystem on the drive, even if the drive is physically fine. Of course, the drive could be dead anyway. You can try those diagnostic CDs I mentioned, they can often read the drive's SMART status, and figure out just what's going on. You could also, once you copy everything you can off 'em, try a filesystem check (checkdisk in windows lingo.) Sometimes just the indexes have been messed up or something, and it's possible to get a bit more data off them after it repairs the errors. Other times, it can actually make things worse, so copy everything you can first. Finally, once you've got all the data you think you can get, and if you want to see what hardware is salvageable, try out the zeroing thing I mentioned.

Good luck, and I hope you get your stuff back!

reinder

Well....

I found a backup of my financial stuff from before the last time I destroyed a file system (that case was deliberate - I wanted a fresh start with SuSE 10), so apart from a spreadsheet of my profit and loss which I made between then and the start of my current troubles, my records are complete.

I've been thinking about the PDF behaviour. The thing about the PDFs is that I didn't do anything to them after copying them to the current drive. The OpenOffice files of the same documents were opened at some stage or another, whether it was simply to look at them or to use them as templates for new invoices. So these were probably touched by the filesystem journal, whereas the PDFs were not. IDE controller breaks, multiple forced reboots, journal kaput, files inaccessible. Almost makes sense even to me.

Also, what with it being reiserfs.... reiser served me well for years. I never touched the default kernels from SuSE, which works well with reiser. But perhaps (K)Ubuntu doesn't? I remembered the other day that Kubuntu hadn't actually been my preferred emergency system - I'd just been using it because it's prettier than Knoppix. But Knoppix is better about automagically mounting drives and may well be nicer to reiser. So I'll try that one today.
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle

reinder

Success! Knoppix recognised my file system and found all my files in good working order. I'm copying some of my art over to the optical drive as I speak.
(I may end up having to do that again as a result of giving the wrong auto-skip instruction, but at least I can verify that the data is present on the drive and in correct working order.)
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle

Xepher

I'll say this, Reiser is an "interesting" file system. When it works, it works great. However, it's got more quirks than just about anything. When it goes bad, it tends to go really bad. A lot of people swear by it, that they've never had problems with it. But the people that do end up with problems usually have extremely bizarre or disastrous problems. I used it for a few months once, actually on the earliest Xepher.net server in my apartment. Worked great for a month, then I did a kernel upgrade, and the new reiser code had slightly changed the way it did some minor operation, and it ate my entire filesystem. It jumped straight into fsck to fix the corruption, and destroyed everything. Granted, that was years ago, and it's certainly improved since then, but... I don't know, I just wouldn't trust my important stuff to it if I had a choice. And now that Hans himself is up on trial for murder, it may not continue getting developed.

reinder

Well I'll be! I hadn't heard about the murder charge. From a little bit of googling, it does look like SuSE/Novell will stop shipping their linux distro with reiserfs as its default file system.

Anyway, I just did the final check on the new drive, loading it again in Kubuntu to see if it showed files correctly now in a system that didn't play as nicely with reiser. Looks like it, and I can now proceed to testing the next drive.

While we're talking about linux: is it normal for some distros to have difficulties working with cable modems? Both Kubuntu and Knoppix refuse to get online when I use them at home, while working perfectly well with the ADSL at my studio. Or is this problem restricted to LiveCD technology, which can't include as many drivers and such as a full distribution? I wouldn't mind switching fully to Ubuntu or Debian (especially if reiserfs is going to fall by the wayside anyway, so any problems Ubuntu might have with it aren't all that relevant) but I do need to be able to get the home PC online.

Mind you, it might just as easily be my home cable modem provider. They're not cooperating well with my iBook either.
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle

Xepher

Well, a lot of cable modems providers can use different technology for authentication. "Normal" ones operate much like I suspect your ADSL. That is to say, you've got a box, one end goes to the wall (be it cable or telephone) and out the other comes regular ethernet. On these schemes, the box/modem handles all the authentication and such. Other providers use things like PPPoE which requires a PPPoE (Point-to-Point Protocol over Ethernet) client on the PC itself. Linux has plenty of support for this, but sometimes it's not included on live cds, or when it is, it isn't activated by default. Gentoo liveCDs, for example, require you to run a helper utility called net-setup if you have an "advanced" connection type such as PPPoE, PPTP, or other more obscure types. A way around this is if you use a router, usually it can handle the connection stuff, and then it will provide regular ethernet to the PC(s.) Check with your ISP to see what they use for set up. Or if you know enough to dig around in the settings on what DOES work, you can probably figure it out there.

reinder

Hmmm... so a shortcut would be getting a router? Me like that idea...
Reinder Dijkhuis
Rogues of Clwyd-Rhan | Waffle