Current Pending Sector count increased, what are the next steps?

[taalas]

Hi,

 

I recently updated my unRAID to the latest version and while enabling the the notification settings I got a report that the Current Pending Sector count of one of my drives is increased (raw value is 64). It might have been increased for a while without me noticing. I checked the other drives and none of them seem to have pending sectors.

 

It is one of the older drives and not flagged as problematic by unRAID yet but I checked the Wiki and saw that it is important to address the problem before another drive shows problems.

 

Before making any mistakes I wanted to make sure what to do:

 

- Should I run a parity check (with or without writing parity)? The last monthly check had no errors...

- Should I buy a new drive straight away to replace this one before trying anything else?

- Should I run any other tests (long SMART) and/or try to re-enable this drive?

 

Any advice would be great.

 

Thanks!

[JonathanM]

1st step would be to make sure your backups are current, so if the array has problems you can recover easily. Current backups remove most of the stress from these issues.

 

2nd would be to get that drive out of the array, and replace it with a tested good drive. The longer that drive is part of the array, the greater the chances of a second drive failing, which could cause data loss. I wouldn't manipulate the array (testing etc.) any more than necessary until you have a known good set of drives with checked parity in place.

[Whaler_99]

I concur on this, just get that drive replaced asap with a new one or one you know is good. Odds are it won't fail any time soon, but better safe than sorry.

[Frank1940]

After you replace this disk, you might want to run three cycle preclear test on this drive.  Watch the 'Reallocated Sector ct.' and 'Current Pending Sector' counts.  Pending Sectors should go to zero and not change. and Reallocated Sectors should increase by the Pending count and then stay stable.  If that happens, you might consider using the drive in some other use.  (There are people who feel that such a drive is actually suitable for use in an unRAID array as the drive is doing exactly what the manufacturer set up the pending and reallocated sector business to do.  It depends on your tolerance of risk taking to determine if you are in that group!) 

[garycase]

As Frank noted, modern drives are DESIGNED to automatically re-allocate defective sectors ... and the process should work exactly as he outlined.    However, what often happens is that when a pending sector is over-written with zeroes (or any other data for that matter), IF the write is successful, the pending status will be cleared instead of reallocating that sector.    This is what you do NOT want to happen, as it means the drive is trying to use a "flaky" sector that can't reliably be read.    That's the reason it was "pending" in the first place -- it had a read issue, and the data couldn't be successfully read, so it couldn't be re-allocated at that time.    Sectors in this category are marked as "pending", and will be reallocated on the next write -- but if that write is successful some SMART firmware simply clears the flag.

 

It's easy to tell if this happens -- if you do the pre-clears (or any other diagnostic that writes to the whole drive, such as WD's Data Lifeguard's "Write Zeroes" function) and the pending count drops but the reallocated doesn't increment by the same amount, then that is what happened.  I would NOT trust such a drive for anything except off-line storage of backups.

 

In any event, I'd definitely replace the drive with a new one, regardless of whether you decide to keep the drive for other uses or simply toss it.

 

 

 

[taalas]

Thank you all for your advice and clarifications!

 

I will replace the drive asap. I have read up on this, haven't done it before though. It seems to be very straightforward. My old drives are still ReiserFS and I understand that the new drive will be ReiserFS too. Is that a problem at all? Or have all problems with this FS been resolved?

 

Also, if I replace the old 2TB drive with a 3TB drive, will the FS be automatically enlarged after reconstruction?

 

Cheers

[garycase]

Reiser isn't an issue -- and yes, the file system will automatically be expanded to take advantage of the larger drive.

 

[taalas]

excellent, thanks alot

[taalas]

Just a quick heads-up: I purchased a replacement drive, pre-cleared it and just started the reconstruction which currently runs at about 120MB/s...everything went very smoothly so far...

[skyhawk]

Generally would these be covered under warranty? Since it's a fault but working as expected, will most companies replace these?

[Frank1940]

Generally would these be covered under warranty? Since it's a fault but working as expected, will most companies replace these?

 

In my experience, the drives are not inspected upon their return under warranty.  (The turn-around time is simply too short.)  What they do after the fact is a matter of conjecture.  BUT, what they probably do is track the number of returns from each person.  You could run into an issue if your return pattern falls outside of what they consider 'normal'.  (As a woild, hypothetical example, if you return three or four drives always in the last month of the  warranty period!)

[garycase]

Agree => I've NEVER had an RMA return challenged.  You simply go to the manufacturer's web site; generate an RMA, and return the drive per the instructions.    A few manufacturer's require that you include the output from their diagnostics; but most don't bother these days.

 

As long as the drive's still under warranty, you can simply return it and receive a replacement.  Note, however, that most replacements are NOT new drives ... they're "refurbished" units (likely returns that have been repaired).  But they'll be warranted for the remaining length of the warranty on your original drive, so you don't lose any warranty coverage as a result of the swap.