Hmm, only 40MB here - although it's the default document root for apache I think, although I just haven't got around to setting that up and changing it since upgrading* the OS a few weeks ago.
cheers
Jules
I think my only "in-use" drive failure was a WD drive - but then that was probably 15 years ago now... :-)
cheers
Jules
ah, you haven't got a mysql database full of important stuff then? or some web servers? Or indeed log files goind back years that you want to do stats on?
By and large, all data that is essentially MACHINE rather than HUMAN generated, lives in /var by default, though not by decree, its true.
watch out for /proc and /dev though.
I've had two failures, both in old 10GB disks which I didn't own from new. One was in a system I built for my mum, which had the foresight to start giving SMART warnings so I replaced it without any data loss. The other was in a small low power system I built as a router/firewall for another family member. I configured the disk to spin down after 5 minutes of non-use, and after a year, it head-crashed.
The others I mentioned before failed before I even managed to copy data on to them, so I count them as DoA.
I use loads of disks at work, and so obviously had some failures there too. Most common is after a power cycle, as the disks mostly all run
24x7.
Until the day your disk controller wipes out the data.
As long as they arent running on compromised hardware.
TBH the best answer is raided on one machine and rsynced to a second. Probably remotely.
speed is not often an issue with personal machines: thats what disk caching is for.
The message from "Dave Plowman (News)" contains these words:
Hard Disks are cheap -- but not necessarily reliable.
Don't trust any of them. I have a mirrored RAID array in my main computer which means, in theory, that when one disk fails you can just take it out (it's even in a removeable tray) and slot another one in with no loss of data.
I've had five yers' use from a couple of Maxtors in that. Not disks with a great reputation, but no problems and 5 years is more than one can really expect to get out of them, so replacement is merely a precaution. I've just replaced them with WD RE3s which have a 5-year warranty.
Backup is to a DROBO which is a RAID-type network attached storage device with a couple of 1TB Samsungs.
And I've a couple of big Seagates as USB external hard drives which can be taken away offsite.
You'll get good and bad examples among the products of all makers. Don't rely on any of them!
The Natural Philosopher wibbled on Wednesday 24 March 2010 17:40
Why should it any more than a hardware controller. I have had a hardware unit drop an entire array of 8 SCSI disks for no good reason. I had to work out the raid structure and superblock size and write a C program to reconstruct it directly from the 8 disks in the end. Linux MD IME is a good deal better at not making a pigs ear and assisted recovery is an easier option than what I had to do.
Irrellevant. If I had compromised hardware, it could just as easily wipe out my data on a hardware RAID unit or a SAN or a NAS.
That's a good idea whatever system you use - I have this setup at home, using my old fileserver with its old slow disks to act as a dump area.
It's an issue with mine - still have to load the cache.
So you just want a disk image then. Done periodically. As well as a proper backup of the data.
Things like norton ghost, will do it, there are free ones if you don't have a copy.
Its probably best to do it to an external USB drive and just swap the drives to do a recovery. Just make sure you buy the correct external drive so its either sata or IDE inside as required.
I've always changed the location of mysql and apache data - it seemed more 'user' than 'machine' data, and I normally set up a /data dir* for user data to live under with relevant directories off that anyway (because quite often I have multiple accounts wanting access to the same data, and I hate having user-contributed data off any home dirs and all mixed up with the 'hidden' dirs that contain user-specific system config).
As for machine logs, I don't tend to keep them beyond a few weeks (not for anything hidden behind a firewall, anyway), although I do sometimes take snapshots of them just prior to any major hardware or software changes so I can look back if needed.
cheers
Jules
Yes, in addition to that I've certainly had a lot that have landed in my lap as DOA or faulty in some way - usually spindle bearing or logic failures (I've not seen that many genuine head crashes). I've had a few drives go bad just while sitting on the shelf (or fail immediately on power-up, anyway). That WD one's the only* one I can think of that's died on me in-use, though.
Yeah, startup's a bad time for disks... the less they can be power-cycled the better, I think.
cheers
Jules
There is method in my madness..
ON a RAID. all the disks are in play all the time.
If the RAID controler goes down there is a better than average chance it will take all with it.
Whether the RAID is hardware of software,is not really an issue.
If you have a simple twin disk system, with one disk spun down., idle only used fr backup, its EXTREMELY unlikely it will get taken out when the main disk gets a hit.
Classic is some kind of power failure and head crash.
Its horses for courses, of course.
A redundant RAID array witt UPS backup is a decent enough beast IF your data is fast moving and totally invaluable. I've used that on a huge mail server serving many people. No point in backing up mail when 90% is transient anyway.
WE were chasing zero data loss and 98% uptime. And we got it.
But here, in single user desktop land, what I am most concerned is not losing days of WORK. I want a backup of everything I do. I am less concerned about losing the operating system and programs. They can be replaced.
twin synched disks works best here. The secondary disk gets very little load. And the option to put it on another machine is very attractive.
I have had a hardware
Oh, no doubt.
I am as much questioning te validity of RAID istelf, in the context of disk failure ion a small single user machine.
I too have had a RAID array totally fail on me. Fortnateely it was in bringing the aforesaid machine up. WE simply demanded our money back, and bought a different solution.
I see RAID as a solution to a different problem than disk failure. Its more about uptime really, and especially in the context of fast moving data.
My point exactly.
The only defense against bad hardware is redundant hardware. A single controller in a single machine is not good. If its only sporadically driving a backup disk, its a lot better. Best is a separate machine and disk controller and disk.
Sadly that IS my main system as well :-)
So its twin disked ..I keep no data of any importance on the dekstop machine. They are too vulnerable. and they get hammered.
well, its seldom my bottleneck
I've used Clonezilla to make a disk images and clone disks for backup purposes. Not the most polished user interface I've seen but it's free.
Yup RAID mirroring...
Something like a Promise RAID controller card would do nicely (I would avoid going for the integrated mobo versions, since you can't guarantee being able to mount the RAID set on a new controller later if the motherboard fails). At least if you can take the card with you it solves that problem.
Had a recent failure of a drive at a clients on a small server we had built for them about seven years ago (machine is on 24/7 but only gets a light workload). The RAID software started to complain about drive errors on one mechanism, so swapped the drive for another, and it just rebuilt the set from the good drive. Total down time about five mins (these were not hot swap drives)>
Connor got taken over by Segate about 14 - 15 years ago. (there are very few drive companies left now).
(a) Any old crap, it's only a PC. (b) Not anything made by Hitachi, for truly they are s**te. (c) You're acting as if you have a choice.
Generally retailers advertise many brands but sell only a few. FWIW Samsung have lasted well for me. The IBM/Hitachi Deathstars lived up their name. WD have also been OK.
I met some people from Seagate at a conference, I wouldn't have bought whelks from any of them.
Depends on how it fails - but there is no reason why it would do wide scale damage to the data content on the drives. Also worth remembering that the controller failure is far less likely that drive failure.
Power failure and controlled shutdown is easy to fix. A head crash only usually affects one drive at a time (unless you drop the whole computer!)
While they can - its very easy to underestimate the amount of work required to get everything re-installed again *and configured the way you want* from scratch.
Rather than synching the second disk, using it to house a multigenerational backup set would give even better protection - the ability to do a full system recovery or just replace a corrupted file as it was 5 versions ago if required.
Indeed its not a replacement for backup.
With entry level NAS solutions so cheap now, that is easy to arrange.
I've had more Western Digitals than any other brand. One went duff shortly after installation, but I got a replacement for free. My most recent purchase was a 250gb WD. But Fujitsus and Seagates have also lasted a long time. My oldest drives must be VERY old, because their capacities are measured in MB not GB! They still work, though.
I make sure I make a frequent True Image backup onto DVD, since DVDs are as cheap as chips these days. My *data* I back up IMMEDIATELY after any updates. That is, when I've finished a session in Visual Basic and have made changes to my current project, I'll WinZip the changed files and whack them on to a floppy AND a USB stick. Then, every few days I'll burn the whole project folder to a CD, also dead cheap. Only problem: I have to keep buying new shelf storage to keep all the backup CDs/DVDs in! It's a shame that no-one has found a use for retired CDs. I once experimented using them as cat scarers for my lawn, but the cats still crapped on it.
However, with all these backups of my actual data (i.e. the truly unreplaceable parts if a disk should fail), I don't really care any more if a hard drive fails. It is much easier to manage nowadays, with the cheapness of media and the reliability of the burn software (I use ImgBurn freeware). Get a bad burn? Just whack another blank in and do it again. We're talking here about pennies per CD/DVD.
MM
atch.
I did that, and lost two months of data when the system crashed due to the backups failing because it ran out of disk space to store all the revisions. And the advertised fix, didn't.
If I want a revision, I'll create it manually
If youre still running windows, just stick a linux live cd in and boot up, and copy the entire disc across.
NT
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.