The hard disk is in the freezer - wish me luck!

So my file server went from working perfectly to "disk IO errors" in the space of a day or two.

I had nightly backups, but when your disk crashes you find what things you forgot to backup. Usually stuff you added since the backups started. Or maybe a MySQL database in a folder which you didn't think to include.

I have had some success in getting stuff off it, but I want the operating system back (Ubuntu). It was set up how I liked it.

So I read about putting the drive in the freezer. So what the heck ... things can't get much worse. If the disk freezes (heh) I've still got most files on backup. I'll let you know how it goes. :slight_smile:

So I read about putting the drive in the freezer

Be sure to keep it dry before using

Have you tried to mount the disk as a secondary disk first? Saved GB's that way.

Yes, good luck, let us know if that trick works.

robtillaart:
Be sure to keep it dry before using

It's vacuum sealed for freshness. :wink:

robtillaart:
Have you tried to mount the disk as a secondary disk first? Saved GB's that way.

I'm booting off a boot CD. The objective is for the drive to stay up long enough to make a copy of the entire thing onto a good drive.

Three things...

  1. An old boss had success running a failing drive belly up. Bear in mind this very likely resulted in a complete failure# a few hours later.

  2. Over the years I've had a total of three harddrive failures. Every one of the failures was the result of a failing power supply.

  3. Good luck!

# No longer spun up.

Last week I took my computer in for a backlight repair, it came out with a new BL, new screen, new keyboard and a new main hard disk. Talk about going into hospital for an appendix and coming out in a box.

I'm happy to report that my backup procedure was mostly good for my data and I used the opportunity to reinstall the OS and make a fresh start. There were a few things I forgot so I bought a HD docking station and retrieved them, next day the disk failed totally.


Rob

There's been times when I wished I had used a RAID. I've got one drive now that may have to go to a data recovery house.

I'm trying now, things seem a bit inconclusive. The drive appears mounted OK, but attempts to do a sector copy fail in the sense that it finished a minute later "all done!".

I'm trying this:

sudo dd if=/dev/sda of=/dev/sdb conv=noerror,sync

Supposedly that should copy everything, skipping bad blocks but leaving gaps on the destination where they were. It actually copied about 900 mB out of 1 Tb.

Since that didn't work I reformatted the destination and am trying to do a recursive directory copy instead. That seems to work slightly better, but when it hits a bad block there seems to be about a 5 minute pause (possibly while the hard disk tries hard to recover) and then it presses on. So, maybe if I can get past the bad part, I will get most files back.

GoForSmoke:
There's been times when I wished I had used a RAID. I've got one drive now that may have to go to a data recovery house.

I think next time some sort of RAID structure is warranted. The ideal is for, if one drive fails, the other one can be used immediately. I had two drives, just didn't have the operating system (ie. bootable) on the second one.

I can report this: Putting the disk in the freezer overnight doesn't seem to have done any harm. So things aren't any worse than yesterday.

The RAID structure I was thinking of is 3 or 4 drives where everything is duplicated spread out. A bit less redundancy? If one quits the rest keep going and you replace it. The drives don't have to match even.

I have the drives but too much on them to duplicate more than 10%. I run them as externals. XP burped and the 1.5T won't talk on the 2 big partitions, soft error on NTSC, delayed write failed ---- funny but I had delayed write disabled and hadn't done anything but started to READ a file. GD M$, I should have known better than connect when not running Linux! So that makes it MY fault! There's a lot of video I don't want to lose there.

I've tried a few different NAS solutions over the past few years; the best open-source solution I found was FreeNAS:

I had such a system set up with 800 GB of space (via two 400 GB IDE drives); I was going to expand it, but things started to go flaky with one of the drives, so I needed a better solution. I ended up going with a DLink DNS-321:

http://www.dlink.com/products/?pid=666

I dropped a couple of 4 TB WD Caviar Greens in, set it up as RAID 1 with Smart monitoring, and have been "pleased as punch" with it since. I use it mainly as a large file server for more junk than I'll ever have time to look it. It also serves as a backup area for the workstations in the home.

In addition, for stuff I download (datasheets, web pages, software, etc) - I also back that stuff up to a 750 GB drive in addition to the file server. I rarely use CDs or DVDs any longer for storage; hard drives (and flash drives for smaller amounts of data) are just a cheaper (and fairly robust) solution overall.

I think DVDs are no longer a viable backup medium. I have about 600 of them for my photos and while I won't ditch them (not yet anyway) I won't be burning any more. I think triple (or even more) redundancy with hard drives it a better option these days.


Rob

I've had more than few CD-R's (out of well over 1000) and a very few DVD-R's (out of many 100's) that didn't last more than a few years without a few coming up with bad spots. If I -really- care about the data I RAR the set then make 25-30% PAR2's and still burn 2 copies.

HD's eventually get screwed up or fail. A net cloud is nice as long as the server(s) holds it.

I kind of wonder about flash in USB sticks and SD/TF chips. I gobble them up on specials but bytes for the buck they are not the cheapest way to store data.

I was up on Amazon ordering media 6 hours ago. Getting a 100-spindle of DVD-R's for just under $20 and Super Saver Free Shipping if I can go over $25 with other SSFS items, I looked at USB Flash drives and hey, all the prices are cut and most have free shipping anyway. I grabbed an 8G for $8 to complete the order (budget says that's all I should do) and I noticed with aching heart:
32G for $26 and 64G for $50

I got a 32G for $26 I think it was last month but the 64G.... plug my ears and tie me to the mast!

the thing to keep in mind about flash is it does eventually fade. they say 10 years, they also say so many cycles which has been blown out of the water. I dont know, but I do know that I have a 128 meg compact flash card that I haven't used since year 2000, and its still loaded up with the Crue and some colecovision games from when I used it in my Cassiopa.

Personally what I want is a 1TB (hell quarter TB) tape streamer for under 99 bucks, I have a small pile of quick 20 and quick 80 tapes from the early 90's that still work fine, as long as you store them in an reasonable environment.

Try changing the block size for dd, that might help.

Backing up a Linux server is easier than people think, it just needs some personal discipline and a good automated bash script.
For Ubuntu every MySQL database will fall under /usr/lib/mysql, www data under /var/www, conf files /etc/apache2 /etc/php5 /etc/mysql.. and so on. Keep the standard locations and you're good to go.

Nowadays not having a RAID system on your home server/NAS..
Well even if you're cheap just go ahead and setup a software RAID partition.

What about ddrescue Ddrescue - GNU Project - Free Software Foundation (FSF) instead of dd?

With regard to flash: some colleagues did high througput / endurance testing with raid arrays combined out of flash drives. They figured that these drives are less reliable in their settings than mechanical drives. The reason was that the number of firmware issues was still way to high. Things may have changed since last year but "not mechanical" does not necessarily mean "more reliable".

With regard to "fading within 10 years": last year I tried to read an old harddrive (~500 MB) and found no controller that would be willing to talk with it any more. I needed to dig out a very old PC running at 100 MHz that was finally capable to copy the data to a 3 GB hard drive (its controller would not talk with anything larger). Then I could connect the 3 GB drive to a modern controller and backup the data. So no matter how long the retention times are: once you do not have suitable controllers or interfaces anymore the retention time is basically irrelevant.

I tried ddrescue, basically it sat there doing nothing, or put it this way, it seems to just cycle through the entire disk once every 40 minutes or so.

I see the foolishness of my ways now. I should have had a bootable disk set up, so whichever one failed, the other would boot immediately. All the installed stuff like MySQL, etc.

As it is, I thought "oh well, I'll just re-install Ubuntu and copy the files back". Wrong! First, the Ubuntu disks are somewhere in The Mess. NVM, downloaded a new one. But then it says "this version is no longer supported" so things like "apt-get update" just fail. So I can't grab the utilities I was relying on (eg. Samba, MySQL etc.).

I think you are right - some sort of RAID dual disk drive. Plus testing to make sure it works.

Our local techie who we use for repairs once told me that most hard drive failures are caused by the circuit board / components on the drive.

He says they often recover data as follows :

Find an identical model drive.
Strip both, and from the broken drive and put the actual hard drive disks into the good drive ( the good drive is no longer a good drive, but if the data on the bad drive is worth more than the cost of sacrificing a drive ... )
Extract the data.

just my 2cents worth

There were data recovery houses back since the 80's that would go as far as opening a drive and moving the platters (if other methods failed) to another drive shell -but- they do it in a clean room and it ain't cheap!

A word of advice from a sysadmin, always.. but ALWAYS install LTS releases.