Failure prediction? (or health assessment)

I have a number of hard drives in a PC acting as a home file server, they range from brand new 2TB SATA to old 250GB PATA with over 40,000 hours runtime on them. I would like to thoroughly test each drive to make sure they are still in good operating condition and suitable for continued use. I know many failures cannot be predicted or prevented, but I'd like to know what tools/tests would be best to help me determine if any of these drives are in need of replacement (and possibly which ones will most likely need to be replaced in the near future). I am relatively familiar with the tools often recommended here (MHDD, Victoria, HDDScan, etc), and I use them on occasion when I suspect a drive is already failing, I also run HDD Sentinel on the server to monitor changes in the SMART attributes 24/7. What do you guys think would be a good series/sequence of tests to confirm that each drive is still in good enough condition for use? I was thinking about running each drive through SpinRite level 3+, then an MHDD scan (with remap on and erase delays for the blank drives?), then run the SMART self-tests (short/extended/conveyance) and check the results and attributes. Does this seem like a good plan? Is it overkill? Could I get by with just one or two of these tests?

Checking the SMART with sentinel is pretty "pro"active as your going to get with predicting HDD failure.

Forget about spinrite..

*This will delete all the data*Make an image or clone it- back it up*

To refresh the disk plug in to your test machine start mhdd. Do a ERASE then just normal SCAN to verify that all sectors are healthy.
You can run a few SMART tests at your pleasure but they wont fix anything.

That is about it.

If you want to be really love your data use some sort of raid system. or just backup what you need

If you get any ERRORS then I suggest you replace the drive or use it on low priority.

4,5 years of life are enough... backup and replace or don't complain later.

ppumkin wrote:Forget about spinrite..

Can I ask, why do you suggest erasing the drive instead of using SpinRite? I was hoping to skip the additional backup and restore steps that would require, and SpinRite allows me to test reads and writes to every sector without affecting the data.

BlackST wrote:4,5 years of life are enough... backup and replace or don't complain later.

If I had the funds to replace all the drives that were older than a certain date, I would love to do that. However I'm short on storage space as it is and I need as many drives operational as possible before I consider buying new ones.

I guess you could use spinrite- it reads the data and writes it back and verifies it.
HDDScan for windows verify function is similar- it just does not rewrite the data back to the drive.
But usually if something goes wrong in spinrite the results are horrible.

Besides - you said thoroughly - My method is pretty thorough. And in most cases if you want to do something properly you are going to have to break a sweat and put your back into it.

At the end of the day- when the drive fails your data will be lost. If you cant afford new drive you wont afford data recovery. So your doomed either way. Its just a matter of time.

I would advise you make a mirror raid of 2 drives and store your utmost critical private stuff on there.
On the rest? movies / tv series and MP3's - honestly what else do you need so many terabytes for anyway?

Failure prediction? (or health assessment)

Failure prediction? (or health assessment)

Re: Failure prediction? (or health assessment)

Re: Failure prediction? (or health assessment)

Re: Failure prediction? (or health assessment)

Re: Failure prediction? (or health assessment)