bit-tech.net

Old Today, 00:33   #1
Behemoth
Timelord in training
 
Behemoth's Avatar
 
Join Date: Nov 2001
Location: Redruth/Weston-Super-Mare
Posts: 1,766
Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.
Bad sectors

Just found I have a bad sector on one of my 2 TB Greens which I was going to put in a NAS project. If I were to use this drive would it cause any errors or would I be better off not trusting it at all ?

I actually thought the drive was newer than it was but it appears to be 5 years old so I've at least had my monies worth if the drive is about to pop off to the big HDD heaven in the sky.
__________________
|AMD FX 4100|8 Gig of Ram|GTX 660|Antec P100|
Behemoth is offline   Reply With Quote
Old Today, 00:57   #2
bulldogjeff
The modding head is firmly back on.
 
bulldogjeff's Avatar
 
Join Date: Mar 2010
Location: Gliding through cyber space
Posts: 7,894
bulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyanbulldogjeff is a Super Spamming Saiyan
It's not totally the end of the world. When you defrag a drive if windows finds any bad sectors it will put the to the end of the drive and just wont use them. The drive will carry on working as normal.
If you are running the HDDs stripped they will be fine because you will have back up should one go down.
bulldogjeff is online now   Reply With Quote
Old Today, 01:03   #3
PocketDemon
I *am* a Dremel
 
PocketDemon's Avatar
 
Join Date: Jul 2010
Location: Sheffield, UK
Posts: 1,889
PocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming Saiyan
The only way to know for sure is with hindsight....

Well, atm, it 'could' be that there was a bad sector from new that was only picked up when you ran some test or other, that it's a temp fault (ie an EEC read error caused by a time out) that could be corrected with a low level format or that it's a new hard fault with the drive.

Now, whilst you could have a look at the smart data using something like CrystalDiskinfo & see if there's bunches of them, what's actually important is how this changes over time...

...so the odd additional one isn't critical - well, it could just be an odd bit that's not as robust as the rest or, as they're not hermetically sealed (other than the He ones), it could just be d.t. a bit of dust having gotten in or... - but if they start increasing fairly rapidly then obviously there's problems.

The only way to know is in retrospect though - it could fail tomorrow for a completely different reason or you could replace it with a brand new drive that fails very quickly; hence the general recommendation on this board to back stuff up.
PocketDemon is offline   Reply With Quote
Old Today, 01:58   #4
Deders
I Mod, Therefore I Own
 
Deders's Avatar
 
Join Date: Nov 2010
Location: Here and there, mostly there
Posts: 3,767
Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.Deders is definitely a rep cheat.
Quote:
Originally Posted by bulldogjeff View Post
It's not totally the end of the world. When you defrag a drive if windows finds any bad sectors it will put the to the end of the drive and just wont use them. The drive will carry on working as normal.
If you are running the HDDs stripped they will be fine because you will have back up should one go down.
If you are worried about bad sectors best to do a full disk scan before a defrag i would have thought?
__________________
i5 750@3800 - MSI P55 GD65 - Asus GTX780 DirectCU II OC 3GB @1215 & 7000 - 8GB Gskill RipjawsX DDR3 @1800 8-9-8 - Corsair HX850i - Nanoxia Deep Silence One - X-FI Xtreme Gamer - Samsung 830 128GB - Spinpoint F3 1TB - WD RE4 2TB - WD Caviar Blue 640 - Dell U2913WM Ultrawide - Logitech G19 - Corsair M65 - Thermaltake Contac 29 with Noctua NF-F12 PWM.
Deders is offline   Reply With Quote
Old Today, 10:02   #5
Gareth Halfacree
WIIGII!
bit-tech Staff
 
Gareth Halfacree's Avatar
 
Join Date: Dec 2007
Location: Bradford, UK
Posts: 4,810
Gareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming Saiyan
Quote:
Originally Posted by bulldogjeff View Post
When you defrag a drive if windows finds any bad sectors it will put the to the end of the drive and just wont use them.
This is absolutely, positively untrue. A bad sector is a failed part of the hard drive. The bad sector is located at a fixed place on the platter; Windows can't move it anywhere, defrag or no. When a sector is marked as bad, it is no longer used for storage; this isn't a Windows thing, it's a hard drive thing.

I repeat for clarity: defragmenting a drive, on Windows or any other operating system, will not move bad sectors around. It can't.

Quote:
Originally Posted by bulldogjeff View Post
The drive will carry on working as normal.
This is true, except that there's a reason the hard drive has developed a bad sector: it's failing. If one bad sector appears, more will follow. When you start seeing bad sectors, it's time to make sure your backups are up-to-date and functional and buy a new drive.
Quote:
Originally Posted by bulldogjeff View Post
If you are running the HDDs stripped they will be fine because you will have back up should one go down.
Dangerously wrong on two levels: RAID is not a backup, it's redundancy for availability; RAID 0 (striping) alternates between disks for writes, with each disk having only unique data - meaning that if one disk in a RAID 0 array dies, you lose all the data across both disks. RAID 1 (mirroring) will survive the death of a single disk, but is still not a backup: it won't protect against any other form of damage or corruption (such as accidentally deleting a file, getting infected by a virus, having a piece of software corrupt a file as it writes it to the disks, and so forth.)

Basically, the '0' in 'RAID 0' refers to your chances of getting your data back after something goes wrong: zero.
__________________
Author, Raspberry Pi User Guide Third Edition, 21 Brilliant Projects for the Raspberry Pi and more | gareth.halfacree.co.uk | twitter | keybase.io
bit-tech news correspondent, Custom PC columnist, other things to other people
I'm a filthy freelancer! Hire me!
Gareth Halfacree is online now   Reply With Quote
Old Today, 12:26   #6
Behemoth
Timelord in training
 
Behemoth's Avatar
 
Join Date: Nov 2001
Location: Redruth/Weston-Super-Mare
Posts: 1,766
Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.Behemoth is definitely a rep cheat.
I've done a full scan over night and it has found now more, doing a scan on the other drive too as thats around the same age.

I'm likely now going to over the next few months start selling off excess kit and then spend the money on 4 x 3 TB drives (Nas spec) rather than WD Greens. Get then another external USB 3 Drive to make backups and keep it off site.
__________________
|AMD FX 4100|8 Gig of Ram|GTX 660|Antec P100|
Behemoth is offline   Reply With Quote
Old Today, 14:49   #7
PocketDemon
I *am* a Dremel
 
PocketDemon's Avatar
 
Join Date: Jul 2010
Location: Sheffield, UK
Posts: 1,889
PocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming Saiyan
Before i start, it 'may' obviously be the case that bulldogjeff has simply gotten the wrong end of the stick about stuff - however, whilst i know that i can be direct when it comes to commenting if someone's saying nonsense, i suspect that two major parts (1st & 3rd ones) may be down to a typo & a lack of clarity respectively...

...so, whilst most of what you've written is correct Gareth (hence the rep as i'd just read his comments differently, but you're right that they really aren't full answers), even for me then i think you've possibly taken the most wrong interpretations from what was actually written rather than looking at what was likely to have been meant.

Quote:
Originally Posted by Gareth Halfacree View Post
This is absolutely, positively untrue. A bad sector is a failed part of the hard drive. The bad sector is located at a fixed place on the platter; Windows can't move it anywhere, defrag or no. When a sector is marked as bad, it is no longer used for storage; this isn't a Windows thing, it's a hard drive thing.

I repeat for clarity: defragmenting a drive, on Windows or any other operating system, will not move bad sectors around. It can't.
Whilst i agree with what you've written on a physical sense (there's an alt argument on a logical sense, but it's not worth going into)...

...there was clearly a typo in bulldogjeff's comment to some degree - so, giving the benefit of the doubt, it perhaps have been more useful to correct that, since -

"...if windows finds any bad sectors it will put the DATA to the end of the drive and just wont use them."

- would be reasonably true. Okay, spare sectors aren't necessarily the end of the drive, but it's a okay way to envisage things.

Naturally, though, if he'd instead meant "...if windows finds any bad sectors it will put theM to the end of the drive and just wont use them." then 'attacking' the validity would have been perfectly reasonable.


The one clear thing to note with his comment is that defragmention could only trigger sector reallocation if the s/w was actually trying to move data either from or to a sector that was found to be faulty - so it really isn't a good way to check drives as a good defragger will neither be attempting to move all of the data on the drive nor attempting to read or write from the entire drive.

So, whilst it's true that it 'can' trigger it, you're actually more likely for errors to be picked up by just using the drive...

...whereas a defined check process or full format would be proper ways to get the drive to check things.

Quote:
Originally Posted by Gareth Halfacree View Post
This is true, except that there's a reason the hard drive has developed a bad sector: it's failing. If one bad sector appears, more will follow. When you start seeing bad sectors, it's time to make sure your backups are up-to-date and functional and buy a new drive.
Whilst semantically you're correct here - in that, over time, more bad sectors will always appear (even if the drive fails completely for some other reason, eventually it will be effected by magnetism & cosmic rays & whatnot) & everything is arguably always failing...

...a getting a bad sector (or even a few) really doesn't automatically mean that the drive is imminently about to fail.

it's a bit like having a cough & looking at a medical dictionary - & whilst you might have plague, you're more likely to just have a cold or a touch of bronchitis or something - though obviously we're all inextricably heading to ultimate failure & the grave.

So, whilst it 'could' be, for example, that the heads are out of alignment & are spasmodically touching the surface or that there's some significant foreign object in there or the controller's failing or something else that could be catastrophic...

...it's more likely that either, as nothing manufactured is truly homogenous, there's just some parts of the drive that will wear out more quickly // or there's a time out error that's not actually a fault with the drive's surface (hence a reason why green drives aren't recommended for raid cards) // or...

indeed, given manufacturing not being 100%, i'd be surprised if at least a significant minority of all new drives arrive with failed sectors (that may be completely invisible to the user - as the smart data comes zeroed) - if not a majority (i can't find any figures to verify this).


Now, obviously, much as some people do get plague & die but will always die for some reason, with data it's sensible to have at least a backup as, whether it's down to failing sectors or whatever, at some point the drive will fail...

...& naturally i also talked about this in my original post, so i am in agreement with you on this point.

Quote:
Originally Posted by Gareth Halfacree View Post
Dangerously wrong on two levels: RAID is not a backup, it's redundancy for availability; RAID 0 (striping) alternates between disks for writes, with each disk having only unique data - meaning that if one disk in a RAID 0 array dies, you lose all the data across both disks. RAID 1 (mirroring) will survive the death of a single disk, but is still not a backup: it won't protect against any other form of damage or corruption (such as accidentally deleting a file, getting infected by a virus, having a piece of software corrupt a file as it writes it to the disks, and so forth.)

Basically, the '0' in 'RAID 0' refers to your chances of getting your data back after something goes wrong: zero.
Everything you've written is completely true, however perhaps it would have been reasonable to assume that bulldogjeff was actually talking about higher raid levels, in the context of the OP using a NAS - since his comment would clearly make no sense to read otherwise?

(obviously, it could alternatively be that he's not understood how R0 works)

Now, obviously you've ignored the possibility that anyone could want to do this, so your answer is incomplete.

That said, obviously a R5/6/whatever array still wouldn't be backup (it's redundancy of course), but it's still a striped array.
PocketDemon is offline   Reply With Quote
Old Today, 15:13   #8
Gareth Halfacree
WIIGII!
bit-tech Staff
 
Gareth Halfacree's Avatar
 
Join Date: Dec 2007
Location: Bradford, UK
Posts: 4,810
Gareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming Saiyan
Quote:
Originally Posted by PocketDemon View Post
Whilst semantically you're correct here - in that, over time, more bad sectors will always appear (even if the drive fails completely for some other reason, eventually it will be effected by magnetism & cosmic rays & whatnot) & everything is arguably always failing...
And if you (or anyone else) wants to trust that the bad sector on a hard drive won't immediately be joined by several more and herald the impending death of the drive, you're more than welcome to do that. I would, however, point to Backblaze's research into the matter which concluded that non-zero reallocated sector count and pending sector count figures "indicate impending disk drive failure."
Quote:
Originally Posted by PocketDemon View Post
...a getting a bad sector (or even a few) really doesn't automatically mean that the drive is imminently about to fail.
I didn't say imminently - and the drive doesn't have to fail completely for it to be a problem. Sure, if a bad sector appears on an empty bit of the drive and is detected and marked as bad, that's of no consequence. But if the bad sector appears right in the middle of your only copy of VeryImportant.file? Whoopsie. (Again with the "backup, backup, backup" mantra.)
Quote:
Originally Posted by PocketDemon View Post
it's a bit like having a cough & looking at a medical dictionary - & whilst you might have plague, you're more likely to just have a cold or a touch of bronchitis or something - though obviously we're all inextricably heading to ultimate failure & the grave.
Except, in this case, it's easy to stave off death: buy a replacement drive. Bosh: your data (which is what we care about here, not the drive) is safe once more. (So long as you have backups, again, 'cos the new drive could die right out of the gate due to a manufacturing defect - darn bathtub curve...)
Quote:
Originally Posted by PocketDemon View Post
Everything you've written is completely true, however perhaps it would have been reasonable to assume that bulldogjeff was actually talking about higher raid levels, in the context of the OP using a NAS - since his comment would clearly make no sense to read otherwise?
If someone says "striped," they mean "RAID 0"; if they're talking about higher RAID levels like RAID 5 (still not a backup) they'd say "striped with parity," the latter being what protects it from the death of a disk - not the striping.
Quote:
Originally Posted by PocketDemon View Post
(obviously, it could alternatively be that he's not understood how R0 works)
Actually, I assumed that he'd got confused between striping and mirroring.
Quote:
Originally Posted by PocketDemon View Post
Now, obviously you've ignored the possibility that anyone could want to do this, so your answer is incomplete.
Ignored the possibility that anyone could want to do what? I ignored nothing: I merely made sure that anyone reading this thread wouldn't go away thinking "hey, great, I turn striping on in my BIOS and now I'm protected against disk failure and don't need to back up" because that would be a horrendous thing for someone to do. Equally, I don't want anyone to read this thread and think "hey, my drive says it has a bad sector so I'll run Defrag to move it to the end of the drive" because running Defrag on a failing drive (and a drive with bad sectors is failing) has a non-zero chance of destroying your data by moving it off a good sector and onto a previously-undetected bad sector.

(Oh, and to correct a piece of your post that could mislead others: if there was data on a bad sector, that data has gone for good. Running Defrag will not, as you suggested in your alternative interpretation of bulldogjeff's post, move the data anywhere: it'll try, but what it gets back from the read operations on the bad sector will be corrupt - which is how it knows the sector is bad, after all. This is why we back up!)
__________________
Author, Raspberry Pi User Guide Third Edition, 21 Brilliant Projects for the Raspberry Pi and more | gareth.halfacree.co.uk | twitter | keybase.io
bit-tech news correspondent, Custom PC columnist, other things to other people
I'm a filthy freelancer! Hire me!

Last edited by Gareth Halfacree; Today at 15:21.
Gareth Halfacree is online now   Reply With Quote
Old Today, 15:24   #9
Corky42
I Mod, Therefore I Own
 
Join Date: Oct 2012
Posts: 3,593
Corky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming SaiyanCorky42 is a Super Spamming Saiyan
Hope I'm not wading in with my size 12's but i didn't think defrag software had the ability to detect bad sectors, or to attempt the relocation of the data contained within, isn't that the job of disk checking software (scandsk).
Corky42 is offline   Reply With Quote
Old Today, 15:25   #10
Craig_T
Minimodder
 
Join Date: Aug 2014
Posts: 35
Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.
This might be a noob post, but hirens boot cd has a few tools on that 'repair' bad sectors, I am not sure how true this is and if it works.. It may or may not be worth a try and I am sure someone of the above will comment on this
Craig_T is offline   Reply With Quote
Old Today, 15:37   #11
PocketDemon
I *am* a Dremel
 
PocketDemon's Avatar
 
Join Date: Jul 2010
Location: Sheffield, UK
Posts: 1,889
PocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming Saiyan
Quote:
Originally Posted by Corky42 View Post
Hope I'm not wading in with my size 12's but i didn't think defrag software had the ability to detect bad sectors, or to attempt the relocation of the data contained within, isn't that the job of disk checking software (scandsk).
Very briefly, if there's any read/write then the EEC checking or a time out that it perceives as a sector error (as it may or may not be) or... ...then it will automatically attempt to correct for them.

Well, if it didn't then we'd all be having to run semi-constant disk checks just in case.

As said though, with a defrag process this will only be with sectors that the data's moved from or to - which, with a good defragger, shouldn't be huge amounts of data unless absolutely required as it's pointless, from a file access time point of view, to randomly rearrange everything.

Quote:
Originally Posted by Craig_T View Post
This might be a noob post, but hirens boot cd has a few tools on that 'repair' bad sectors, I am not sure how true this is and if it works.. It may or may not be worth a try and I am sure someone of the above will comment on this
Not all sector errors are actual failures of the disk - so, for example, a TLER (Time-Limited Error Recovery), or equivalent on non-WD drives failure that was d.t. to some random glitch 'could' cause a sector to be wrongly marked as faulty.

What something like Hiren's *should* be doing is getting the drive to retest these sectors - so if they are actually faulty they should stay so, but if not then they'd be reset.

Obviously though, 'if' the process were not accurate or the sector failures were a weakness under certain conditions or over time, it 'could' lead to faulty sectors wrongly being assigned as working - which would increase the risk of losing data.

That's not to say don't use something like it but that it may not have consistently useful results.

Last edited by PocketDemon; Today at 15:56.
PocketDemon is offline   Reply With Quote
Old Today, 15:44   #12
PocketDemon
I *am* a Dremel
 
PocketDemon's Avatar
 
Join Date: Jul 2010
Location: Sheffield, UK
Posts: 1,889
PocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming Saiyan
strange double post
PocketDemon is offline   Reply With Quote
Old Today, 15:57   #13
Craig_T
Minimodder
 
Join Date: Aug 2014
Posts: 35
Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.Craig_T is the Cheesecake. Relix smiles down upon them.
So, essentially what you are saying if I understand it right is... It's worth a try, in case they aren't genuinely 'faulty' as such because it will reset them. If they are broke, it will just tell you where to go and you've not caused anymore damage?
Craig_T is offline   Reply With Quote
Old Today, 16:01   #14
PocketDemon
I *am* a Dremel
 
PocketDemon's Avatar
 
Join Date: Jul 2010
Location: Sheffield, UK
Posts: 1,889
PocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming Saiyan
Quote:
Originally Posted by PocketDemon View Post
Very briefly, if there's any read/write then the EEC checking or a time out that it perceives as a sector error (as it may or may not be) or... ...then it will automatically attempt to correct for them.

Well, if it didn't then we'd all be having to run semi-constant disk checks just in case.

As said though, with a defrag process this will only be with sectors that the data's moved from or to - which, with a good defragger, shouldn't be huge amounts of data unless absolutely required as it's pointless, from a file access time point of view, to randomly rearrange everything.
Quote:
Originally Posted by Craig_T View Post
This might be a noob post, but hirens boot cd has a few tools on that 'repair' bad sectors, I am not sure how true this is and if it works.. It may or may not be worth a try and I am sure someone of the above will comment on this
Not all sector errors are actual failures of the disk - so a TLER (Time-Limited Error Recovery), or equivalent on non-WD drives, could cause a sector to be wrongly marked as faulty.

What something like Hiren's *should* be doing is getting the drive to retest these sectors - so if they are actually faulty they should stay so, but if not then they'd be reset.

Obviously though, 'if' the process were not accurate or the sector failures were a weakness under certain conditions or over time, it 'could' lead to faulty sectors wrongly being assigned as working - which would increase the risk of losing data.

That's not to say don't use something like it but that it may not have consistently useful results.

Quote:
Originally Posted by Gareth Halfacree View Post
lots of stuff
Clearly we took different interpretations from what was written - i wasn't attacking you, simply saying that i'd looked at the mistakes differently.


As to odd faulty sectors, you are obviously welcome to replace every drive with any fault sectors - but it's really not necessary.

Though, arguably, if 'one failed sector is too many' it would also make it impossible for you to ever buy any drives as there is a significant likelihood of there already being faulty ones within the manufacturing process that have already been marked as such - with the smart data then being set to zero afterwards.
PocketDemon is offline   Reply With Quote
Old Today, 16:08   #15
Gareth Halfacree
WIIGII!
bit-tech Staff
 
Gareth Halfacree's Avatar
 
Join Date: Dec 2007
Location: Bradford, UK
Posts: 4,810
Gareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming Saiyan
Quote:
Originally Posted by PocketDemon View Post
As to odd faulty sectors, you are obviously welcome to replace every drive with any fault sectors - but it's really not necessary.
I beg to differ. I don't mess around with my data: it's literally how I keep a roof over my head. If I lost my work files, I'd be in serious trouble.

I don't know if you read Backblaze's report on how reallocated sectors are a strong indicator for pending drive failure, but if you did and chose to ignore it allow me to add Google's white-paper on the matter (PDF warning) to your reading list. To quote Wikipedia's synopsis:
Quote:
Originally Posted by Wikipedia
In the 60 days following the first uncorrectable error on a drive (SMART attribute 0xC6 or 198) detected as a result of an offline scan, the drive was, on average, 39 times more likely to fail than a similar drive for which no such error occurred. First errors in reallocations, offline reallocations (SMART attributes 0xC4 and 0x05 or 196 and 5) and probational counts (SMART attribute 0xC5 or 197) were also strongly correlated to higher probabilities of failure.
Granted, 56 per cent of failed drives in the study failed without SMART warnings, but when a SMART warning does show up I pay attention. D'you know what? I've never lost data.
__________________
Author, Raspberry Pi User Guide Third Edition, 21 Brilliant Projects for the Raspberry Pi and more | gareth.halfacree.co.uk | twitter | keybase.io
bit-tech news correspondent, Custom PC columnist, other things to other people
I'm a filthy freelancer! Hire me!
Gareth Halfacree is online now   Reply With Quote
Old Today, 20:47   #16
PocketDemon
I *am* a Dremel
 
PocketDemon's Avatar
 
Join Date: Jul 2010
Location: Sheffield, UK
Posts: 1,889
PocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming Saiyan
Quote:
Originally Posted by Gareth Halfacree View Post
I beg to differ. I don't mess around with my data: it's literally how I keep a roof over my head. If I lost my work files, I'd be in serious trouble.

I don't know if you read Backblaze's report on how reallocated sectors are a strong indicator for pending drive failure, but if you did and chose to ignore it allow me to add Google's white-paper on the matter (PDF warning) to your reading list. To quote Wikipedia's synopsis: Granted, 56 per cent of failed drives in the study failed without SMART warnings, but when a SMART warning does show up I pay attention. D'you know what? I've never lost data.
i had to head out earlier - hence not replying in full at the time.


A. BackBlaze.

Firstly, they're solely looking at drives in a server environment - so is this reasonable for a home one?

They then state -

"From experience, we have found the following 5 SMART metrics indicate impending disk drive failure:
•SMART 5 – Reallocated_Sector_Count.
•SMART 187 – Reported_Uncorrectable_Errors.
•SMART 188 – Command_Timeout.
•SMART 197 – Current_Pending_Sector_Count.
•SMART 198 – Offline_Uncorrectable."


- however, there are no %ages given for 4 of them, &, for the one they do -



- then it's complete nonsense, since, according to the data, they've got failure rates of >100% which, unless they're bursting on fire & damaging other drives or something (which they're not claiming), is clearly impossible.

The only thing that i can assume is that they've multiplied all of the %ages by 10x inadvertently - otherwise we have to write the results off as completely worthless.


Now, if this is the case, this instead would mean that, even in the 65-120 error range, there's only something like an average 19% chance that the drive would fail...

...so, even if this were a key metric where there was general *causation* for any drive usage, as opposed to a correlation in a server environment, >80% of the drives would be replaced needlessly.

* * * * * * *

B. Then moving onto the Google document.

Now, that's one side of it, which, if what you were saying was actually backed up by the evidence, seems to show a pretty extreme causation; however, without attempting to cover everything, there are some clear issues here even if we solely focus upon scan errors (which is where the 39x comes from in the document itself).

So, running through some easy points -

(a) Firstly, the 39x is meaningless here since, as you note yourself, we've largely been talking about the reallocation count thus far in the thread.

Well, when looking at that, it says that "The average impact on AFR appears to be between a factor of 3-6x." - though the same arguments below will apply whichever of the two metrics we use.


(b) Secondly, we're looking at drives that were made prior to 2007 - so we're ignoring any advancements in reliability since then.

***Now, the OP's drive is ~5 years old so the data may not be valid at all.***

More generally though, we're not imagining that, on this forum, someone's come across a cache of unused 8-9+ year old drives that they're proposing to use, are we?


(c) Thirdly, even if the data were still 100% valid, you have to look at what these numbers actually mean, otherwise it's complete nonsense - & 39x a 0.000000001% chance vs 39x a 2% chance have a totally different impact upon risk of course.

So, as a couple of easy comments -

(i) A little over 70% of all of the drives with a scan error survived the first 8 months after getting one... So that's <30% that actually needed replacing for some reason.

(ii) The most significant proportion of drives, in actual numbers, that fail having had a scan error are those less than 1 months old.

As noted, this may be the tail end of the infant mortality curve - having already lost x% of drives in manufacturer testing - however by including this in the overall data, this gives a misleading impression of the failure rate...

This is then followed by those between 3-4 years old (the middle graph in figure 8 is misleading as it bands all drives from 2 years to 44 months together).

So, if you have a drive that's younger than a month & within the 3-4 years range then there might be an increased risk (see (d) below)... ...whereas if you have a drive that's any other age then it's reduced.

***Now, the OP's drive is ~5 years old so does not fall into these higher risk ranges.***


(d) & fourthly, not having data on what other issues the drives which did fail had or conditions doesn't prove that there's any causation between scan errors & failure rates.

Well, because the results have been separated out there is no indication as to how the chance for any age of drive as a result of scan errors has been impacted by any other factor...

So, for example, as the results show that drives running below 30C have a higher rate of failure until Y4 - when it's then switches to be drives above 35C (increasing with temp), it 'could' be the case that all of the early failures are actually down to low temps - whilst the latter ones are high ones...

...& that scan errors (& other smart errors) that lead to actual failure are primarily a result of running the drives at temperature extremes.

Now, i am not saying that this is literally the case that temps are the cause - but as 56% of the drives had no smart errors in the 4 main areas then the graphs don't have to align; it 'could' simply be that the scan errors are a symptom that a reasonable %age of the drives show when temps are the cause...

...but, all of the way through, the report explicitly talks about there 'appearing to be a correlation' - not that 'there *is* a causal link' - & it is a complete misreading of the data to assert anything different.

See www.tylervigen.com for examples.

* * * * * * *

in short, the claim you've quoted of a 39x increase means absolutely nothing in terms of the reallocated sector count (as it's a different value), OR for a modern drive, OR an unknown aged drive in or before 2007, OR that there was anything more than an apparent correlation (correlation does not equal causation)...

...so what you have quoted is completely meaningless in terms of the actual likelihood of anyone's drive actually failing in the next 8 months *solely* as the result of a reallocation error today.


Now, even *if* the data in both cases happened, by pure coincidence, to actually show a true *causation* for any drives in any environment (again assuming that BackBlaze have multiplied their %ages by a factor of 10), you're looking at replacing drives on the basis of a (A) <20% or (B) <30% chance of failure within 8 months.

So, on that basis (which isn't actually shown), if you've got the cash to replace 100% of the drives when the data shows that >80% won't fail within the year that the errors occur or >70% will last at least 8 months then how lucky you are.


Now, that's not suggesting that anyone should be actively reckless with things, but whilst it would probably be better to err on the safe side & not actively choose to pair two drives with scan errors as being a main drive & its backup, imho it would not be a wholly unreasonable risk for one of them to be if there isn't all the money in the world.

Last edited by PocketDemon; Today at 21:19.
PocketDemon is offline   Reply With Quote
Old Today, 21:02   #17
PocketDemon
I *am* a Dremel
 
PocketDemon's Avatar
 
Join Date: Jul 2010
Location: Sheffield, UK
Posts: 1,889
PocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming SaiyanPocketDemon is a Super Spamming Saiyan
Quote:
Originally Posted by Craig_T View Post
So, essentially what you are saying if I understand it right is... It's worth a try, in case they aren't genuinely 'faulty' as such because it will reset them. If they are broke, it will just tell you where to go and you've not caused anymore damage?
Answering this separately from the long reply, it's completely a judgment call.

As said, it could correct errors that were caused by some random one off temporary issue - but it could potentially reset genuinely faulty sectors to being non-faulty which could then potentially lead to data loss.

(so a small chance of a small chance of future data loss)

Now, whilst the safest option would not be to try it, in truth it's not something i've thought of trying for maybe 10-12 years or so tbh & i had some successes, back in the day, with a similar premised (though DOS based) program...

...so i'm neither saying don't try it nor actively recommending that you do.
PocketDemon is offline   Reply With Quote
Old Today, 22:25   #18
Gareth Halfacree
WIIGII!
bit-tech Staff
 
Gareth Halfacree's Avatar
 
Join Date: Dec 2007
Location: Bradford, UK
Posts: 4,810
Gareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming SaiyanGareth Halfacree is a Super Spamming Saiyan
Quote:
Originally Posted by PocketDemon View Post
[snip lots]
First, I want to thank you for taking the time to write all that out. I've read it through, and you make very good points about the studies and their relevance. I also accept your more general point that replacing a hard drive when it has only one or two bad sectors may be poor economy.

I'll keep doing it, mind, because I value my data (and the time I would spend restoring things from backup) more than the £50 I spend every four or five years replacing drives that might not necessarily need replacing at that exact moment. (And I don't waste the old ones, either: I've got a 1TB Seagate Recertified drive (i.e. one which had already failed once and been fixed by Seagate) which has a couple of bad sectors, and I use it as a scratch drive and temporary storage in a USB caddy.)

You have a different approach, one which saves you money but exposes you to increased risk of data loss, albeit arguably only slightly. That's perfectly valid, and clearly works for you. I won't be adopting it, but there's no reason I should - any more than you should adopt my more expensive and risk-averse approach. Horses for courses, innit?
__________________
Author, Raspberry Pi User Guide Third Edition, 21 Brilliant Projects for the Raspberry Pi and more | gareth.halfacree.co.uk | twitter | keybase.io
bit-tech news correspondent, Custom PC columnist, other things to other people
I'm a filthy freelancer! Hire me!
Gareth Halfacree is online now   Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 23:46.
Powered by: vBulletin Version 3
Copyright ©2000 - 2015, Jelsoft Enterprises Ltd.