HDD current pending sector back to 0 after random pass

sdauth · Guru Joined: 19 Sep 2018 Posts: 569 Location: Ásgarðr

Hi,

A couple of days ago, one of my old (2012) 3.5" HDD smart status shown 4 "current pending sector". I quickly replaced it with a new one. Today, I decided to check again the old disk with a short smart test, which quicky failed at the same LBA.
Then, I ran a full /dev/random pass to erase it and put in my dead HDD box..
But with much surprise, after the random pass, I noticed the smart value for current pending sector was back to 0. So I ran a full smart test this time, and it finished without error.
What could explain this behaviour ? Or maybe is it expected ? I always assumed that when that value was going up (as with "Reallocated Sector Count or Uncorrectable Error Count) then the HDD was heading to the grave..

mike155 · Posted: Sat Feb 19, 2022 9:20 pm Post subject:

I can only guess.

A HDD can remap faulty sectors. In order to remap a faulty sector, the drive must read the sector. After that, it can write the data to a spare sector. If the drive can't read a sector to be remapped, it will increase the number of current pending sectors.

If you send a write command to the sector that needs to be remapped, the HDD can skip reading the faulty sector. It will write the new contents directly to the spare sector. After that, it will decrease the number of current pending sectors.

I'm not surprised the number of current pending sectors went down to zero after you copied /dev/random to your disk.

See: https://harddrivegeek.com/current-pending-sector-count/, section "Can You Fix/Lower Your Pending Sectors Count?"

sdauth · Guru Joined: 19 Sep 2018 Posts: 569 Location: Ásgarðr

Thanks for the explanation, appreciated. From your link : "Pending sectors are the precursor to reallocated sectors which can be a strong indicator of a dead hard drive on the horizon."
For now that value is at 0 so all good but from now on, I will only use that disk for cold storage and monitor closely smart values when I power it on. :wink:

At least, I can still use it to store some files.

figueroa · Posted: Sun Feb 20, 2022 5:28 am Post subject:

I have an over 10 years uptime 500 GB WDC WD5000AAKS-00UU3A0 that has been reporting

Hu · Moderator Joined: 06 Mar 2007 Posts: 21624

eccerr0r · Posted: Tue Mar 01, 2022 3:03 am Post subject:

As long as it still has spares in the same replacement zone as the pending sectors, a write to that failed sector will cause a remap and clear that pending sector. Copying /dev/zero or anything (including a sequential rebuild in a RAID) to a hard disk to erase or rewrite it, by that virtue, will clear out every pending sector and attempt to remap them.

Usually there shouldn't be a SMART error on BIOS boot or smartd for just having a few bad sectors. Only if you have a SMART condition alert where WORST < THRESHOLD will a SMART warning flag. Bad sectors don't count... yet...

I have a 2TB disk with 150 pending sectors, 15 offline uncorrectables, 364 reallocations already occurred in the past, and 408 rellocated sectors so far. This is called a slow moving train wreck as I keep on forcing reallocates and new ones pop up. The reallocations already tipped off SMART that this disk is on the verge of failing as it set WORST to 1 for this field. TBH, this disk is effectively dead already. It's unfortunate, I could use another 2TB of disk space...
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

figueroa · Posted: Tue Mar 01, 2022 4:42 am Post subject:

The only reason that I haven't yet acted on Hu's suggestion is time to do it, and the need to be careful as this is a remote server. Although that drive has 4 pending and 2 offline uncorrectable sectors, the reallocated sector count is zero. On the other hand, your (eccerr0r) 2TB drive should be demoted to toy status.
_________________
Andy Figueroa
hp pavilion hpe h8-1260t/2AB5; spinning rust x3
i7-2600 @ 3.40GHz; 16 gb; Radeon HD 7570
amd64/23.0/split-usr/desktop (stable), OpenRC, -systemd -pulseaudio -uefi

eccerr0r · Posted: Tue Mar 01, 2022 4:58 am Post subject:

It's tough to fix sectors when the drive is mounted read/write, really need to make sure the filesystem knows about what you're doing in case it has a write pending from an open file to that sector. On the other hand the data is already lost, it might be enough to just try to find out what file (or directory or metadata) contains the bad sector...
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

sdauth · Guru Joined: 19 Sep 2018 Posts: 569 Location: Ásgarðr

So I just powered it on today since I need more storage for some backups. I made a short smart test (via USB adapter) and it seems ok.
Would you use it ?

pjp · Administrator Joined: 16 Apr 2002 Posts: 20067

NeddySeagoon · Posted: Fri Jul 29, 2022 5:13 pm Post subject:

sdauth,

sdauth · Guru Joined: 19 Sep 2018 Posts: 569 Location: Ásgarðr

NeddySeagoon · Posted: Fri Jul 29, 2022 6:51 pm Post subject:

sdauth,

A non zero Current_Pending_Sector count is a count of the sectors that the drive has tried to read but failed.
It would therefore relocate them, if only it could read ream.
In short, the drive cannot read its own writing.
Its grounds for a warranty return unless the warranty has expired.

Its no quite that simple. Every time the drive goes to read a sector, it also gets a 'measure of difficulty' associated with the read.
At some threshold, difficult to read sectors are reallocated and the originals abandoned, with the data still in them.

Sometimes a read fails and the sector added to the Current_Pending_Sector count. Then a subsequent read works and its below the relocation threshold, so the sector is removed from the Current_Pending_Sector list without being relocated. This is what happened to you.

The Current_Pending_Sector count is only the sectors the drive has tried to read and failed.
There may be many more that its not tried to read yet, so are unknown.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

pjp · Administrator Joined: 16 Apr 2002 Posts: 20067

sdauth · Guru Joined: 19 Sep 2018 Posts: 569 Location: Ásgarðr

@pjp
Yeah I'm going to buy a new one very soon.

@NeddySeagoon
The HDD is currently being filled up to the max. (1/3 done currently), so once it is done, I assume that if I generate a hash for each file, it will 100% triggers a read failure (or most likely multiple) at some point right ?
All of this is really a bad idea haha and I should just buy a new drive right now

At least, I've been warned.

Thanks again for the exhaustive explanation.

NeddySeagoon · Posted: Fri Jul 29, 2022 8:57 pm Post subject:

sdauth,

As long as you build the hash from the sources and don't save it is the failed drive, maybe.
All hash functions have a finite length. That means that there must be collisions in the hash space.
(Several different inputs will have the same hash value)

You will notice bad blocks in your media collection, the application will stop playback.
A hash is not required.

How will you know if the media collection is bad or the hash is bad if there is a mismatch in future?
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

sdauth · Guru Joined: 19 Sep 2018 Posts: 569 Location: Ásgarðr

I finally replaced the disk, I mounted the failing drive read-only (plugged in on a SATA port directly instead of using my USB3 adapter) and I've been able to transfer everything to the new one without any issue. :wink:

No read error during the transfer so everything looks good.