HDD Smart Error

[nims076]

Hi,

I currently have 8.2.0 p1, this has 5 x 3tb drives set up as RaidZ1.

I get the following SMART error reported:

The following warning/error was logged by the smartd daemon:
Device: /dev/ada0, 5 Currently unreadable (pending) sectors

From what I have read, I need to replace this HDD and not write to the NAS at all.
Does this drive have to be replaced? Can I somehow repair it? A low level format etc?

My query is, what is the best way to do this?

I am thinking the process it to:

1. Click 'replace' in the GUI for the faulty drive and the shutdown and remove the drive

2. Its a 3tb Hitachi so I have to use software like HD Tunes to run a test against it while connect to a windows system
Is it possible to test the drive from the NAS itself?

3. Drive is under warranty so if there are errors found then do an advanced RMA

4. Insert new drive when it arrives and from GUI click 'Replace disk' and the select 'Replace'

5. Let the drive resilver and then run a scrub.

Please let me know if this the best way forward

Thanks

 

[Stephens]

You need to read the Documentation (6.3.11) on the drive replacement procedure. Since you only have a RAIDZ1, I'd think seriously about whether it's worth risking your whole array waiting for a RMA drive as opposed to just buying one and putting it in now. If another drive fails while you wait for the new drive, bye-bye array. After the resilver, you can do a scrub if you want, but you things are likely OK once the resilver completes (note: completes, not says 100%).

 

[nims076]

I am happy to leave it offline whilst I wait for a replacement drive but I will probably take your advice and buy one anyway so I always have a spare.

Is it best to remove it and do the error tests using a windows pc with HD tune?

Do you know if a low level format will work or do I definitely have to replace the drive?

Thanks

 

[peterh]

"low-level-format" is a thing of the past. If a modern drive has errors reported with smart - replace !

 

[Stephens]

The specifics aren't exact, but here's what I think normally happens. You ask for a block of data. The system goes out and reads "HDD1" for the block. It gets a read error because it can't read it. SMART takes note. The drive may retry some indeterminate number of times. Eventually, the sector will get reallocated. ZFS takes note. ZFS reconstitutes the data from your parity on the other drives. ZFS resilvers this data onto HDD1 (in a good sector).

Generally, when drives start going bad, you don't find ways to continue using them in critical applications. You COULD, but do you really want to? Go to a command shell and type "zpool status -v" (assuming you only have on pool). What's it say? Also, what does SMART say about reallocated sectors?

I remember the days of low level formatting since drives would lose their magnetic properties over time and reformatting would refresh them. There were even application that would do a nondestructive refresh of the whole disk (reading and writing every sector), though those worked at the logical level (refreshed sectors, not the low level tracks). I honestly doubt I've done that since the days of RLL drives. I can't speak with authority as to if that works at all these days. I've occasionally seen mention of low level formatting to recalculate the bad sectors table and remapping table, etc., but all I can keep thinking is. "It's $150." (for 3TB... in my case, it's $100 for 2TB drives). In other words, it's not worth my time to try to figure it out or to spend time trying to beat a dead horse (dying drive). I might rotate the drive to non-critical duty (even as a non-critical backup drive, or a drive on a testbed system), but it's time in my NAS would be over.

 

[nims076]

Ok thanks for the info guys. It was more to save me the hassle and wait times of replacing the drive, fortunately the drive is still under warranty.
Point taken, I will replace the drive.
Last question, can I replace the 3tb with a 4tb? From what I have read I believe I can but is it the same process as replacing a like for like?
Thanks

 

[Stephens]

Yes, but it will only use 3TB of the 4TB drives. If you eventually replace the other 3TB drives with 4TB drives, FreeNAS/ZFS will expand the size of the pool upon replacement of the last drive.