Slow hard drive, but SMART looks OK

[Larry Sabo]

After I've imaged off a suspect drive I like to run it through HDD Regenerator and let it tell me how many delays, bad sectors, re-allocated sectors, etc. the drive has before I decide to continue with that drive. At least every sector gets checked and it beats on the drive a bit so if things are going to happen I want them to happen now.

I don't see any benefit to running HDD Regenerator. My understanding is, it reads from then re-writes data to the drive to "refresh" it and if it can't do so successfully, flags the sectors as bad. Doing so accelerates its demise. You could read-verify scan it with HDDScan to assess whether it has slow or bad sectors without beating it to death. However, it takes forever to do that and new drives are just too cheap to waste the time and tie up your computer doing so. Personally, I wouldn't trust a drive that has been assessed or "refreshed" with HDD Regenerator.

 

[Diggs]

I don't see any benefit to running HDD Regenerator. My understanding is, it reads from then re-writes data to the drive to "refresh" it and if it can't do so successfully, flags the sectors as bad. Doing so accelerates its demise. You could read-verify scan it with HDDScan to assess whether it has slow or bad sectors without beating it to death. However, it takes forever to do that and new drives are just too cheap to waste the time and tie up your computer doing so. Personally, I wouldn't trust a drive that has been assessed or "refreshed" with HDD Regenerator.

Hmmm.... Should have been more explicit. I don't usually fix or "refresh" the drive but it reports info I don't get other places. ...and if I do run it in "fix" mode the little "beating" it takes is mild compared to the real world. If it can't take it it shouldn't be used. I'd prefer it die during my beating than the customer's beating.

 

[lcoughey]

Hmmm.... Should have been more explicit. I don't usually fix or "refresh" the drive but it reports info I don't get other places. ...and if I do run it in "fix" mode the little "beating" it takes is mild compared to the real world. If it can't take it it shouldn't be used. I'd prefer it die during my beating than the customer's beating.

If it is what you'd prefer, go for it. I'll stick to MHDD (DOS) or Victoria (Windows) which are free.

 

[Porthos]

it takes forever to do that and new drives are just too cheap to waste the time and tie up your computer doing so.

This 100%

 

[MudRock]

I've seen WD Greens fail this way. Make a full image or at minimum data back up first, then run through some of what is suggested here. My guesses would be in this order: Drive -> Driver -> O/S -> Controller -> Motherboard

I've had some seriously evil deaths from Greens. Seagate, when they die, they die bad. When a Blue goes, you know its heads and its a fast death. But greens seem to die this evil slow death which you don't realize until nothing is doable, even at data recovery. The only drive I've sent off in the past which was irrecoverable was a Green.

 

[Porthos]

I've had some seriously evil deaths from Greens. Seagate, when they die, they die bad. When a Blue goes, you know its heads and its a fast death. But greens seem to die this evil slow death which you don't realize until nothing is doable, even at data recovery. The only drive I've sent off in the past which was irrecoverable was a Green.

So true.

 

[Markverhyden]

It's taken Fab's 30 minutes to scan the directories on the drive and copying has (finally) commenced.

This decision of which one to do first (capture user's data first or ddrescue/cloning) is always something of a quandry for me. And I've never seen this slow drive with no SMART errors symptom before. Again, thanx for all your input

With delays like that I'd have stopped and jumped to doing an image after 10 minutes or so. Just never know what might happen. Last week a customer's MacMini server choked. Booted from a USB drive, tried accessing but stopped after 10 minutes or so. Even though they had Time Machine backups I still imaged the drive, just in case thing, took about 25 hours for around 80gb.

 

[Haole Boy]

Update.... I made the decision to go ahead and run Fab's first. Customer emphasized that recovering her photos was the primary objective and I did not want to risk having the drive die while doing a full clone and not having captured any of her photos. If this was a case of just wanting to clone the hard drive, I would have run ddrescue first.

Fab's has been running for 17 hours now and is not finished, but it has only reported 2 unreadable files.

To answer a few questions and comments from the thread

- Drive has been pulled from the customer's machine and is connected to my bench machine via an eSATA docking station. Bench machine is dual boot Windows (for Fab's and Drive Snapshot) and Linux Mint (ddrescue)

- re: HDDRegenerator. I do own this program and have used it many times to repair drives so that I can get a more complete image to clone with. But I always debate if I should run HDDRegenerator first and then run ddrescue, or run ddrescue first, then run HDDRegenerator and then run ddrescue again to pick up the cleaned up sectors.

- Temperature of the drive (touching it with my hand) was slightly warmer than room temp after running Fab's for about 30 minutes. A couple hours later it was hot, so I set up a fan to blow on the drive. Cooler to the touch now.

- I had not heard of HDDSuperClone. Just took a look at the webpage for this program. Very interesting, Will have to look into this some more.

Mahalo nui loa for all the responses and suggestions.

Harry Z

 

[Haole Boy]

With delays like that I'd have stopped and jumped to doing an image after 10 minutes or so. Just never know what might happen. Last week a customer's MacMini server choked. Booted from a USB drive, tried accessing but stopped after 10 minutes or so. Even though they had Time Machine backups I still imaged the drive, just in case thing, took about 25 hours for around 80gb.

That's one of my biggest challenges with trying data recovery - having the patience to let the ddrescue (or whatever program) finish.

 

[Diggs]

If it is what you'd prefer, go for it. I'll stick to MHDD (DOS) or Victoria (Windows) which are free.

Appreciate the heads up @lcoughey . Didn't know about Victoria. I need other tools. I go way back with HDD Regenerator but not so much any more.

I find I get a much better FABs backup if I run the Regenerator but it's not my first choice.

 

[lcoughey]

Fabs is more likely to kill the drive and get less data recovered than cloning with ddrescue. But, fabs backup is better than HDD Regenerator which could kill the drive with copying a single sector.

 

[Diggs]

Actually, all the testing I've done is that FABs always got a ton more stuff after HDD-R than by itself. SMART cleans up on the drive. It's not the best way to go about things but it has merit. (...and again - after imaging has been done...)

 

[Markverhyden]

Update.... I made the decision to go ahead and run Fab's first. Customer emphasized that recovering her photos was the primary objective and I did not want to risk having the drive die while doing a full clone and not having captured any of her photos. If this was a case of just wanting to clone the hard drive, I would have run ddrescue first.

Fab's has been running for 17 hours now and is not finished, but it has only reported 2 unreadable files.

To answer a few questions and comments from the thread

- Drive has been pulled from the customer's machine and is connected to my bench machine via an eSATA docking station. Bench machine is dual boot Windows (for Fab's and Drive Snapshot) and Linux Mint (ddrescue)

- re: HDDRegenerator. I do own this program and have used it many times to repair drives so that I can get a more complete image to clone with. But I always debate if I should run HDDRegenerator first and then run ddrescue, or run ddrescue first, then run HDDRegenerator and then run ddrescue again to pick up the cleaned up sectors.

- Temperature of the drive (touching it with my hand) was slightly warmer than room temp after running Fab's for about 30 minutes. A couple hours later it was hot, so I set up a fan to blow on the drive. Cooler to the touch now.

- I had not heard of HDDSuperClone. Just took a look at the webpage for this program. Very interesting, Will have to look into this some more.

Mahalo nui loa for all the responses and suggestions.

Harry Z

The point of imaging first, if it's a bit by bit image, is that it is for all intents and purposes a copy of the patient hard drive. So you should be able to run any tool you can run on a regular HD on an image. Of course things that claim to "repair" bad blocks won't work since they only exist on the old drive. But any other filesystem utility should work. If, after an initial evaluation, there is any doubt I'll clone. Then proceed with the other related tasks.

Personally I've always been leery of these apps that claim to repair a drive with problems. Especially if all they are using is the native hardware and driver layer. The way I see it is, at best nothing will get worse and worst case total drive failure.

 

[fincoder]

I've seen drives with 100% SMART then doing a surface verify suddenly I have thousands of pending sectors

SMART status is good for diagnosing a faulty drive, but not for ruling out a faulty drive.

The SMART statistics only indicate issues that have been previously detected. I use Crystal Disk Info and it can be a quick and non-destructive way to confirm a faulty drive. However if it says Good, a full surface scan test (I use Seatools) would be needed to rule out a faulty HDD.

 

[Diggs]

Personally I've always been leery of these apps that claim to repair a drive with problems. Especially if all they are using is the native hardware and driver layer. The way I see it is, at best nothing will get worse and worst case total drive failure.

From my understanding (and reading the white papers) there is more to it. Usually a drive error starts as a parity bit(s) that's been tripped. The data may be fine and can be read and re-read to determine the best possible pre-state by comparing the data to the parity bit with an algorithm. Only special software takes the time to look in detail at the data, re-examine the bytes and try to recover with intelligence. The OSs just mark as bad, move what they know is good and move on.

...and again - I work residential. If it's grammy and the grandkids and a few recipes then a quick HHD-R is more appropriate. Small business or family books and records mean something else entirely. It's discussed with each customer.

 

[Porthos]

It's discussed with each customer.

This is #1 when dealing with data.

 

[NviGate Systems]

I own HDD Regenerator License and use it time to time. If the data is not critical and client wants a clone, I will often run it to stabilize the drive. This is good for drives that have say 10-40 pending sectors and a full read/verify didn't show evidence of further damage.

It's also good when the one pending sectors happens to be your boot record. Hate those ones.

 

[frase]

I've seen WD Greens fail this way. Make a full image or at minimum data back up first, then run through some of what is suggested here. My guesses would be in this order: Drive -> Driver -> O/S -> Controller -> Motherboard

I've had some seriously evil deaths from Greens. Seagate, when they die, they die bad. When a Blue goes, you know its heads and its a fast death. But greens seem to die this evil slow death which you don't realize until nothing is doable, even at data recovery. The only drive I've sent off in the past which was irrecoverable was a Green.

Greens and Blues are crap even though Blue is my fav colour - I always go for WD Black, seagate are a piece of shite

 

[Diggs]

Greens and Blues are crap even though Blue is my fav colour - I always go for WD Black, seagate are a piece of shite

The new WD Black received yesterday was much lighter and using a different chassis than in the past. Much the same as the Blues. I have more Blacks coming in today. I'll check them out. Hope they didn't cheapen up the Black series.

 

[MudRock]

The new WD Black received yesterday was much lighter and using a different chassis than in the past. Much the same as the Blues. I have more Blacks coming in today. I'll check them out. Hope they didn't cheapen up the Black series.

I'm struggling to keep faith in WD. I have had so few Blues die; I've actually had more Blacks die than Blue (Even though more blues do go out). However, I'm seeing models which are by old standards Greens labelled as Blues now, and Blacks that may perform better but worse failure rate than many Blues.