How to fix Currently unreadable (pending) sectors?

[Ryan Beall]

I'm currently running a raidz2 setup with 6 drives and have had two drives start popping unreadable sectors. The first drive went from 1 to 80 sectors in over 6 hours. I attempted to follow some guides on another google post which were pretty bad and there was limited advice on these Forums that I could track down. I was unable to get the drive to reallocate the pending sectors to un-used backups via my results from the smart tests. About this point, the second drive started popping bad sectors as well so I rushed to get a new drive to keep from losing the whole pool. I replaced the worst drive (80+ bad sectors) and re-silvered with no issues. I still have the second bad drive in the pool and it's now at 40 pending sectors? It has been stable at 40 for about a week now but I'm not really sure where the limit is for being repairable or what my next steps should be.

Is there a step by step guide to get these pending sectors to swap to the backups? Doing long smart tests to find the beginning of the pending sectors and then manually writing over some blocks etc etc takes days of effort with little to no feedback. At 40+ sectors, is it smarter to just offline the drive and format the whole thing and then resilver? I don't have any extra hardware to hook up to the drive so I will have to do everything via freenas if possible. I'm currently on travel and won't be back in country to get a usb to sata adapter etc for a couple more months.

Thoughts, advice?

Thanks!

 

[joeschmuck]

Yea, time to install some good drives. You probably don't understand the physics behind what is happening to the hard drive but basically the surface of the platter is failing or even worse, the head is failing, but it's normally the platter surface. You cannot fix this problem.

You can read about "badblocks" but that isn't the solution and it's destructive.

 

[danb35]

What @joeschmuck said. In the short term, you can remove a disk from the pool and run badblocks on it. That will write to each sector four times, which will force a remap of any bad sectors. That will make your SMART numbers look better, but it won't really fix the problem. The problem is that the disk is beginning to fail, and the way to fix it is to replace the disk.

Edit: After running badblocks, you'd then replace the drive into the pool, which would resilver your data. Your pool would be in a degraded state while running badblocks, and until resilvering was finished, though.