I'm just wondering how aware people are of these issues, and if they do anything to prevent them. Be honest now, this isn't about planning to do these things at some point in the future, but rather about actually doing them, right now.

It doesn't matter what OS you use, that's why I posted it in this forum.

EDIT:
I did not add an "other" option, because I think this covers everything, if you don't think so, say so.

I've got a bunch of rsync scripts that run every night and copy files between my machines. I may lose one machine, but I doubt I'll lose them all. One of the disks is a portable USB drive, so I can easily just grab that and run if I need to.

Both regular backups (via cronjob) and irregular backups (when dealing with "more" important files) to external harddisk(s) and, for some files, to another computer. I don't find harddisk testing utilies practical at this scale, and because of low disk prices I doubt if there's much sense in any scale. It's cheap enough to have several external disks and rsync'ing (or doing something similar) incremental copies is quick. If a disk (when checked) started to look bad, one would have to switch it anyway, and on the other hand keeping an eye on the disk won't prevent a sudden accident, so in the end having up to date backups is the only "good" way to prevent a catastrophe. Of course it won't harm to follow the situation by other means as well, but I think it's just "extra". RAID makes sense in the sense that it may allow one to continue working without interruptions when disk has to be replaced with a new one, but it's again extra cost and work, and in my case doesn't make that big a difference. In a nuclear power plant it might be more important.

I use SMART as well as irregular backups to DVD. I do a smartctl long test once every 3 months or so, and I backup data either when I need more room on the drive (rare), or when I re-install slackware, or immediately for important data.

So far, I've never had data loss from a failing HDD, smartctl helps detect when the drive is getting old, and I just replace it before it fails. This is probably why I've never had a HDD fail on me, I replace it beforehand. I've thought about using RAID (1 in this case), but I don't trust it. Either way it's mostly for servers that need to keep data available at all times, but it can be used so that if one drive fails, the other is still ok. However, if you buy two identical drives, they could very well fail at just about the same time (within hours or days) ... that's why I don't trust it.

So far this scheme is good enough for me.

Also, I posted this poll, because I see a lot of people don't take any precautions and then when the HDD fails, there's hell to pay, and you never get back 100% of the data. I also agree that a good backup scheme is the most important thing here, probably something like rsync to multiple other media. However, for me this is kinda overkill, my data is not that important. Besides, I have older backups that are just as useful as newer backups (they contain almost the same data, I can afford to lose all the new funny pictures and videos and pr0n that I've collected in the mean time, that is not important).

Hi

+1 CronJob/Backup to external HDD

My backups are OTT at the moment because the new system is not fully working and tested so the old system is still in use.

• Old system: custom script using dar to backup to same internal HDD then copy dar backups plus key recovery files (LVM, system build, dar utilities) to DVD with par2-generated error checking and correcting data. Alternate DVDs (even-numbered days of month) stored off-site .
• New system: Bacula to another internal HDD and custom script to copy Bacula backup volumes plus key recovery files (LVM, system build, Bacula utilities) to set of 3 USB HDDs of which one always off site.
• Belt and braces: user data synchronised between desktop and netbook.

RAID 1, regular bare metal backups to 2 different external drives, some offsite backups. I look at SMART once in a blue moon...

Mostly backup to CD/DVD. After my first external HD failed, I luckily had most of my data also on a friend's external. I did buy another HD to store data, however this time I bought an internal WD with an antec external enclosure. I'm never buying anything ever again from Seagate! Unfortunately now I am trying to make copies of what I have on that drive, but it is just too much.

Backing up 300+ GB of music, and other stuff just doesn't seem appealing with DVDs, and dual layers are still pricey unfortunately.

I kinda have mixed feelings about using bluray. I don't care about bluray video, and yes bluray = bad because of all the DRM shit, but I am considering bluray however on strictly backup purposes only. I mean, 22GB (single layer) per disc is rather appealing instead of 4.4GB (single layer) or 8.1GB (dual layer).

It would be nice to have to use less discs, so I guess I'll have to wait for next gen optical media formats: HVD, 3D Optical Storage.

Hi,

PMS- Preventative Maintenance Schedule in a manner & frequency to hopefully show any potential problems. Within this schedule for the system(s), regular maintenance and software scheduled maintenance is performed. Been doing it this way for years.

Subsystem maintenance depends on the system usage. Server maintenance is different than a desktop/laptop. Periodic testing hopefully benefits or shows potential problem areas before any catastrophe. In the last forty or so years some failures were environmentally caused so no fair indicator if PMS prevented or caused it. Some were caused by environmental maintenance people (janitors) who caused the filters to plug thus thermal damages to hdd & PSU at same & different times.

Price & availability of hdd systems does allow one to become lax but habits do win out. Most problems are mechanical, connectors and such. My hdds' seem to last forever! Especially when good hardware is purchased to begin with. KNOCK on WOOD!

I've got some museum systems(circa mid 70s') that still fire up. Loads of single board systems & 80ish systems that still boot.

I'm old school and believe in leaving the systems up if they're going to be used within a period. While at the University, my lab units ran 24/7. Only problems were with systems that had to be cycled a lot.

Obviously I'm not a tree huger but believe one should have the choices to not waste. Al Gore sets the example for me! Flying around in that jet preaching the world needs us to go green. Go nuclear!
What then?

I spin down when necessary.

I don't take any precautions right now, so I didn't vote, but I plan on using an external hard drive if I start backing everything up...

One word SMART

I don't do anything particularly special, but I do tend to keep my personal files "synced" (not with rsync or anything, just manual scp-style copying ) between my laptop and main desktop (the wimpy desktop is my "mad science" machine, it doesn't need anything personal on it, really ).

As for optical media, I don't have any DVD-R/RWs in the house, so no can do on that...would be nice to try though, if I ever get the chance, since AFAIK my laptop is capable of burning DVDs (it's a "SuperMulti" drive to the best of my knowledge ).

I too have consider blu-ray as a means of backup, but the burners cost more, and disks cost a lot per GB of storage, much more than DVDs. And if it is not rewriteable, and you mess up writing the blu-ray, it is a very expensive mistake. Until the prices go down, I will not be using this.

As for Seagate, did the failure happen because of the firmware ? If not, then I don't see any reason to blame them, HDDs do fail, from every company.

Couple of things to add (for personal computers, not servers) ...

I've got SMART set up to email on detecting an out-of-bounds condition because it's hard to keep checking things that are always (hah!) normal. I do scan /var/log files semi-regularly, especially syslog and messages.

A major downside of backing up to DVD is that it slows the system down enough to noticeably impact interactive response, even using nice and ionice. Backing up to USB HDD a) completes in a much shorter time and b) is only noticeable when working interactively after an idle period -- nice/ionice kick in quickly to restore interactive response.

I have a pretty comprehensive system, I'm somewhat paranoid about losing data.

My live copy of all my data is on a RAID 10 system, this gets backed up every 6 hours using rsync and hard links to an internal drive. The internal drive gets mirrored to 2 external drives twice weekly. It's all automated, even down to udev checking the serial numbers of drives plugged in to determine if it's one of the back drives. If it is, then the backup script is run and I get an email when it's done and safe to remove the drive.

External/Offline backups are encrypted so I can keep one offsite.

I also have smartd running that looks for error states every so often