SOLVED Prioritizing drive replacements, failed SMART test vs unreadable sectors

[funwithdickandjane]

Hello all,

I'd had a drive with unreadable sectors for a couple weeks now and just got the replacement drive for it but now I also have a drive that just started reporting it's not capable of SMART self-check. Here's the daily email I get.
Device: /dev/ada5, not capable of SMART self-check
Device: /dev/ada5, Read SMART Self-Test Log Failed
Device: /dev/ada8, 2 Currently unreadable (pending) sectors

Which of these should I replace first?

 

[Jailer]

Without seeing the actual smart output from each drive it's going to be near impossible for anyone to make a sound judgement call.

 

[jgreco]

ada5 is more worrying to me since it suggests a bigger failure is underway there. Drives often develop a few unreadable sectors and it is not a catastrophe, it just means you've lost the ability to read two sectors, which will be remapped at the next write. That *could* be indicative of problems brewing, but the SMART test failing *IS* a sign of badness.

 

[funwithdickandjane]

Thank you. Drive has been replaced and I'm off to RMA another drive :/

 

[jgreco]

I meant to point out that if you start to see an increase in the "2" that this would also indicate a developing problem and that you should get more worried IFF that happens.

Hard drives are devices that rely on incredible levels of precision materials, engineering, and physics to store and retrieve data, and the thing that amazes me is that we can design these devices with such high reliability. Two sectors on an 8TB drive, for example, is a one-in-a-billion failure rate (sectors).

 

[JakeTheWonderdog]

I know this is resolved, but let me point out a couple of things:

"I'd had a drive with unreadable sectors for a couple weeks now and just got the replacement drive..."

If your data is important to you, don't wait until a disk fails to get a replacement / or RMA a disk. You need to have at least one replacement disk in stock and on-site. For larger arrays you may need more than one on site to cover multiple failures or to have one replacement while you wait on a turn-around for an RMA.

Going a couple of weeks before replacing a disk is way too long if it has actually failed or is failing.