HD Diagnostics / How long will it last?

[BassKozz]

I recently built a test-bed system with spare parts I had laying around, and i bought 5.25-inch bay hotswap drive rack for 3.5-inch IDE PATA (http://www.newegg.com/Product/Product.asp?Item=N82E16817119503) and another hotswap drive rack for SATA HD's (http://www.newegg.com/Product/Product.asp?Item=N82E16817123301) (I know these models are on the cheaper side, but they are getting the job done on a budget)...

And I now have a box full of HD's (probably over 20), some open some un-opened, some more then 100gigs some less, some only a few years old some older, most are out-of-warranty. You get the picture, It's clean up time.:D

My goal is to test all the drives...
My plan is to start with the drives manufacturers diagnostic software...
Seagate & Maxtor Drives = SeaTools (http://www.seagate.com/ww/v/index.jsp?locale=en-US&name=SeaTools&vgnextoid=720bd20cacdec010VgnVCM100000dd04090aRCRD)
Western Digital = Data Lifeguard Diagnostics (http://support.wdc.com/download/)
Hitachi = Hitachi Drive Fitness Test (http://www.hgst.com/hdd/support/download.htm)

Besides the manufacturers own diagnostics software can you guys recommend any software to test the drives?
Is there any way to test how much longer the drive will last ? (Life Expectancy)

Thanks in advance for the help,
-BassKozz

[drizzt81]

Is there any way to test how much longer the drive will last ? (Life Expectancy)[/B]
According to Google Lab's recent study on this subject, there are no reliable tools for predicting drive failures.

[BassKozz]

According to Google Lab's recent study on this subject, there are no reliable tools for predicting drive failures.
That's disappointing :(
What do I do with all these old drives?
I don't want to use them (for fear of loosing the data I place on them) especially if I can't somewhat predict when it's going to die.

[drizzt81]

I was in a similar predicament recently. I put most of my older drives into a single machine and created a couple of Linux software raid arrays. This way, if a drive fails, I won't lose data. It is my online backup server (aka. it holds duplicate data).

If this is a `hdd graveyard' machine and you keep the storage utilization at a reasonable level, you can reduce the array sizes as drives fail, not requiring you to purchase replacements.

[BassKozz]

what version of linux did you go with ?

-BassKozz <---- Linux Newb :p

[unhappy_mage]

If you happen to have hardware on the HCL (http://www.sun.com/bigadmin/hcl/) around, you might try Solaris 10 (http://www.sun.com/software/solaris/get.jsp). Then you can use ZFS, which does an admirable job of conglomerating a bunch of disks (in whatever size you happen to have) into one logical volume. It's probably not as easy to set up as Linux would be, but if you're going to take the plunge you might as well go all the way. You can get redundancy easily - either build your zpool out of mirrors, or run "zfs set copies=2 filesystemname".

[BassKozz]

I think Solaris might be outa my league, but I'll give it a try...

As far as linux goes whats the best version to use for testing disks and recovering data?

[drizzt81]

I think Solaris might be outa my league, but I'll give it a try...

As far as linux goes whats the best version to use for testing disks and recovering data?

Doesn't really matter on the distribution as much as on the filesystem you select. I would heed U_M's advice and consider Solaris. I have not yet used it, but ZFS seems like a really smart filesystem. According to Wikipedia, ZFS can detect bit (byte?) errors on the fly. RAID-n only notices when a HDD has failed, but not if data is corrupted.

If you are set on Linux, I like Debian, since that is what I first learned, so I am familiar with it. Ubuntu/ Kubuntu are ``based on'' Debian and may be a little easier. If you are planning to install Linux and want the OS partition mirrored, you will need to do a bit of research (and/ or ask) beforehand. I find that the people in the Linux/BSD/Free Systems (http://hardforum.com/forumdisplay.php?f=96) subforum are very helpful with all things.

[unhappy_mage]

Doesn't really matter on the distribution as much as on the filesystem you select. I would heed U_M's advice and consider Solaris. I have not yet used it, but ZFS seems like a really smart filesystem. According to Wikipedia, ZFS can detect bit (byte?) errors on the fly.

Yes, byte. ZFS uses end-to-end checksumming; when you write a block it stores that block's checksum along with the metadata, and when you read the block back it consults the metadata to make sure the block has survived the slings and arrows of being stored on disk.

Also, FreeBSD now has support for ZFS. If you're looking for something slightly less Unix-y, it might be a good way to go.

[BassKozz]

Time to start playing with Solaris and FreeBSD :D