"I usually never work on the laptop so much as use them as dumb terminals to ssh and browse from, so I have no backup regiment when it comes to laptops at all."
I had to learn that lesson the hard way.
Hard drive data recovery = $2000+
2005-11-07 07:11 am (UTC)
I think I'll just rewrite that one module. :-)
If you have a static/ziplock bag and a silica packet around, put the drive in the bag with the packet, seal it really well then throw it in the freezer overnight. I've done this myself for dozen of co-workers, and I usually get 1-2hrs out of the drive (usually enough to get any files off).
Doesn't always work, but more often than not I get good results with this method.
I'm curious... why does this work?
I think it has something to do with freezing the read/write head back down to a manageable size. when the drive clicks and dies, i think it has something to do with the head being enlarged...freezing it makes it smaller again...the reason it only works for a little while is because it does eventually defrost and grow.
i can't find anything to back that up, but i think i read it somewhere.
2005-11-07 07:25 am (UTC)
From what you've said, chances are the hard drive was that close to dead. If a drive is dying you generally don't want to let it spin down/power down until you've copied the data you want off it. Spinning it up again is often the hardest thing to do.
That said, I wouldn't completely give up on the drive just yet, especially if there's only one thing on it you really really want. Leave the drive alone overnight, and try it again in the morning. And perhaps again tomorrow evening. A period to cool down can help (some people even try putting them in the freezer for a little while -- maybe 15-30 minutes -- but beware of the moisture). If you get it to spin up, then copy the thing you really really want off first, then see what else you can copy. If you don't have space locally, copy stuff over the network, rather than trying to move it to another machine. rsync is your friend.
I've been lucky that way with several drives, including recently with a (four drive) RAID-5 set where two of the drives died "at once". With a lot of patience we managed to make each of them work again for long enough to mirror them to replacement drives. After that one of them was completely dead, never to work again, and the other is pretty marginal.
Finally when you reach the "nothing to lose" point, you might be able to get the drive to spin up by applying power and then giving it a quick spin in the clockwise direction. This is very hard on the drive, so definitely not recommended for normal use, but has been known to help drives with spin up issues get free enough to spin up. Generally you should plan that the spin up you get from it is your last. (The other method which was used in the past was to drop the drive onto a table -- perhaps 2"-3" down onto a flat surface. That's also pretty much a last resort, and was more useful with heavier drives than the 2.5" laptop drives, but it can sometimes bounce things free.)
As someone else mentioned there are data recovery services that might be able to get data off the drive for you (they replace logic boards, motors, etc, to make it go for a bit), but they tend to be very expensive. For a small project doing it a second time is generally more cost effective.
PS: An occassional rsync to some server of the whole drive makes for a pretty easy to do backup for a laptop. Even if it's done manually and only run when you remember it's "been quite a while".
2005-11-07 07:29 am (UTC)
Heh. You've been through this a couple times it sounds like. :-)
with a (four drive) RAID-5 set where two of the drives died "at once".
Hopefully they weren't on the same IDE channel. :-)
2005-11-07 07:35 am (UTC)
The drives were SATA disks, so no they weren't on the same IDE channel. I'm not sure exactly what order they died in, as I've never actually seen the server -- it's on the other side of the world -- and only found out the second disk was dead after asking for the system to be restarted.
And yes, I've been through the "dying disks, quick get the data off" several times, and the "oh dear, the disk is 'dead', what will we do" a few times as well. It's usually in those situations people find the problems with their backup strategy...
"From what you've said, chances are the hard drive was that close to dead. If a drive is dying you generally don't want to let it spin down/power down until you've copied the data you want off it. Spinning it up again is often the hardest thing to do."
Yup yup, I made that mistake once, and *poof* no more re-spinup. It took me *two* hard drive deaths to smarten up. I lost all my data since the 286 days D: I'm never buying WD again! The failed drives were theirs, all within less than a year :\ Also, this RAID-5 biznit is a false sense of security. Nowadays I've got my date in RAID-5, but I should go back to saving stuff onto DVD/hard drives and putting that stuff into a fireproof safe.
Everything edm said, are techniques you can typically use to squeeze another few hours out of a dying drive... after that, it's off to replacing the controller board (you have to match the batch, and it's notoriously hard to do), the motors, and/or read/write heads. I wouldn't trust myself to do teh latter two though.. I'd probably kill it further. I've only replaced the controller board for a few Maxtors (another brand I don't trust!).
2005-11-07 07:53 am (UTC)
When tiffany's laptop HD was making the click of death, I found that if I held it at a wierd sideways angle, it didnt click. Dont know if you could try something like that. Ive heard the freezer trick works.
Or you can be hardcore and just boot from a customized knoppix CD and use a usb keychain as an impromptu storage disk.. :)
2005-11-07 08:08 am (UTC)
Heh. Nice icon.
I usually never work on the laptop so much as use them as dumb terminals to ssh and browse from, so I have no backup regiment when it comes to laptops at all.
Plug for arch / distributed repositories. I guess if it's just one file you're working on, that's probably overkill, versus just remembering to cp or rsync. But for working on stuff and committing offline, then having your offline commits filter back into the main tree when you plug back in, star-merge rules.
Of course, I know you aren't an Arch guy, so I'll just stop there. :)
2005-11-14 02:01 am (UTC)
Is there any real reason to use everything over ssh rather than to run programs locally and then rsync data back later?
2005-11-14 02:03 am (UTC)
Re: mobile computing
More of a bummer is that I have no Windows/IE/Opera to test with for a while.
VMware is your friend. Got four VMs running, sleeping or ready to boot on this box as I type: W2K with IE6+firefox, IE5.5+opera and IE 5, and W2K/AS with SQL Server if I ever need it. There'll be a fifth with XP/IE7 soon, and if I can get Intel OS X up on one and persuade it to run IE/Mac in some arcane and vile emulation mode, I can retire the iMac at the end of my desk too.