HDD issues.

[Spork Schivago]

Re: HDD issues.

It is too bad they just don't remember the bad sectors and skip them the need time you read them so that the drive would still be usable.

Yeah, I believe that's how it's supposed to work. The drive marks a sector bads and tries to move the data to a non-damaged sector on the drive. Or at least, that's what I remember from back in the day. Don't know where I got that information from though, it could of been from college or just an idea my friends and I came up with.

In this case, I kinda need the data in the damaged sectors. One of the sectors contains the hibernation file. When I try copying that file manually, it errors out. I'm gonna restart the PC and see if the drive drops back down to UDMA / 33 mode. Just running ddrescue the second time took longer than the first time.

[stj]

Re: HDD issues.

if you open failed ones you would know.

usually there is an area on the platter with the coating rubbed away like a ring.
i could be where the swapfile was stored, or it could be where a regularly used file resides.
whatever the cause, the head must get worn down too.

i know drive makers write the firmware to scatter data across the drive to try to stop this happening.

however, this can cause issues with smooth reading on slow drives, so the firmware does exactly the opposite and tries to write files in continuous blocks on AV drives designed for PVR's

WD's AV-GREEN series are a good example of this.

[keeney123]

Re: HDD issues.

Linux can do that, it's called badblock remapping.
but as Chaos said, it's only going to get worse - usually at an accelerating rate.

That is interesting because the hard disk just holds magnetism in it. I wonder what the failure analysis reports are on that? Magnetic tape has a life expectancy of 30 to 100 years. I wonder why a hard disk would have a shorter time?

[stj]

Re: HDD issues.

It is too bad they just don't remember the bad sectors and skip them the need time you read them so that the drive would still be usable.

Linux can do that, it's called badblock remapping.
but as Chaos said, it's only going to get worse - usually at an accelerating rate.

[ChaosLegionnaire]

Re: HDD issues.

that wont work and is just pointless on a failing drive because the bad sectors would just keep increasing and increasing. it can no longer store data reliably. what is the point in continuing to use such a drive? russian roulette with your data? no thx and no spinning pun intended.

[keeney123]

Re: HDD issues.

I think there's bad sectors on the drive, not so much the heads, because I can read the other partitions just fine. Maybe some dust or something? I think the reason it drops down is because when it can't read the data, it tries a bit slower. I might be totally wrong though Keeney123. Definitely not a hard drive expert, but I do know a good bit about PCs.

It is too bad they just don't remember the bad sectors and skip them the need time you read them so that the drive would still be usable.

[Spork Schivago]

Re: HDD issues.

I think there's bad sectors on the drive, not so much the heads, because I can read the other partitions just fine. Maybe some dust or something? I think the reason it drops down is because when it can't read the data, it tries a bit slower. I might be totally wrong though Keeney123. Definitely not a hard drive expert, but I do know a good bit about PCs.

[keeney123]

Re: HDD issues.

That is interesting. Do you not think that if it became unstable it would not damage the read/write magnetic head?

[Spork Schivago]

Re: HDD issues.

like i said, most likely the drive is faulty and is unstable at the higher udma speeds so it drops to the lower speed. i had a faulty drive intermittently do that to me before.

Yeah, I agree with you ChaosLegionnaire. I believe that's what's happening.

[ChaosLegionnaire]

Re: HDD issues.

like i said, most likely the drive is faulty and is unstable at the higher udma speeds so it drops to the lower speed. i had a faulty drive intermittently do that to me before.

[Spork Schivago]

Re: HDD issues.

firstly, yes using an esata connection would fall under the definition of using a native sata connection for the drive.

secondly, if u are recovering a drive with multiple partitions and hundreds of megabytes of bad sectors, its probably easier to recover the entire physical disk at once rather than just individual partitions at a time. use another hard disk that is the same size or larger than the faulty hard disk. use /dev/sda /dev/sdb with the -f force switch to recover an entire disk to another disk. the -f switch is to override the safety feature of gddrescue that normally prevents partition or entire disk overwriting due to accidental misuse of path parameters.

for the first pass, run it with the -n switch but WITHOUT the -d direct disc access switch. this is important to speed up the initial rescue and let it run as fast as possible. using direct disc access disables caching which will slow down the initial data rescue.

then for the second pass, if the drive is still alive, run with the -d and -r 1 switch to attempt recovery of the lightly dmged sectors.

after that for the third pass, run with the -d -A -M -r 3 to retry the remaining bad sectors but in a different order (thats what the -M retrim switch does), making some of them readable when read in a different order.

thereafter, use -d -A -R -r 3 for the fourth pass to try all the remaining bad sectors but in reverse order or backwards. sometimes reading a bad sector backwards will make it readable again.

then if u still got bad sectors left, for the fifth pass try -d -A -M -R -r 3 to retry all the remaining bad sectors but in a different reverse order to try to make more bad sectors readable.

if even after all that, there are still bad sectors remaining, now u have no choice but to attempt the extreme deep freeze recovery method that i described earlier.

while the drive is still frozen and cold, reuse the commands i mentioned in the third, fourth and fifth pass to try to grab the data off the bad sectors. hopefully, this will recover all your data.

this should not be happening. make sure the sata cable is not accidentally knocked loose or anything or if the cable is faulty or if the goldfinger contacts on the drive are dirty. could also be due to a faulty circuit board and the drive really living on borrowed time... this is why u use the -n switch!

Thanks for all the help. I believe the reason it dropped down was because of all the errors. Anyway, finished the first pass last night, gonna run the second now. After it finished, it did some stuff, trimming or something like that and the non-recoverable data went all the way down to 32216 kB. I'll try now with:

ddrescue -d -r1 /dev/sdc4 /home/spork/backups/sdc4.img /home/spork/backups/sdc4.logfile

Only reason I didn't try the entire hard drive at first was because I didn't realize it was bad. I saved sdc4 for last. I must of missed ddrescue erroring out on the 5th partition there. When I mount /dev/sdc5, it's just a recyling bin and a System Volume Information (which contains a file called tracking.log). I figured someone partitioned it for user data and they just never used it. sdc4 is my main concern right now.

dmesg shows: configured for UDMA/33

Perhaps this drive is somehow configured that way at the factory or something? Doesn't make sense at all though. Why they'd configure a SATA drive to run sooooooooo slow. It does explain why it took ddrescue soooo long to just copy the 300GB partition though. Since I started ddrescue again, we're now down to 32183 kB, so that's progress! Baby steps I guess!

[ChaosLegionnaire]

Re: HDD issues.

firstly, yes using an esata connection would fall under the definition of using a native sata connection for the drive.

secondly, if u are recovering a drive with multiple partitions and hundreds of megabytes of bad sectors, its probably easier to recover the entire physical disk at once rather than just individual partitions at a time. use another hard disk that is the same size or larger than the faulty hard disk. use /dev/sda /dev/sdb with the -f force switch to recover an entire disk to another disk. the -f switch is to override the safety feature of gddrescue that normally prevents partition or entire disk overwriting due to accidental misuse of path parameters.

for the first pass, run it with the -n switch but WITHOUT the -d direct disc access switch. this is important to speed up the initial rescue and let it run as fast as possible. using direct disc access disables caching which will slow down the initial data rescue.

then for the second pass, if the drive is still alive, run with the -d and -r 1 switch to attempt recovery of the lightly dmged sectors.

after that for the third pass, run with the -d -A -M -r 3 to retry the remaining bad sectors but in a different order (thats what the -M retrim switch does), making some of them readable when read in a different order.

thereafter, use -d -A -R -r 3 for the fourth pass to try all the remaining bad sectors but in reverse order or backwards. sometimes reading a bad sector backwards will make it readable again.

then if u still got bad sectors left, for the fifth pass try -d -A -M -R -r 3 to retry all the remaining bad sectors but in a different reverse order to try to make more bad sectors readable.

if even after all that, there are still bad sectors remaining, now u have no choice but to attempt the extreme deep freeze recovery method that i described earlier.

while the drive is still frozen and cold, reuse the commands i mentioned in the third, fourth and fifth pass to try to grab the data off the bad sectors. hopefully, this will recover all your data.

I know the drive has dropped down to UDMA/33 now. ddrescue is just about done.

this should not be happening. make sure the sata cable is not accidentally knocked loose or anything or if the cable is faulty or if the goldfinger contacts on the drive are dirty. could also be due to a faulty circuit board and the drive really living on borrowed time... this is why u use the -n switch!

[Uranium-235]

Re: HDD issues.

I've had external laptop drives act like they're dying if i'm using an USB extension cable from the voltage drop due to the added wire length and the fact it's DC

usually the freezing trick works cause some electronics on the board is overheating (cause it's shorted of otherwise fried), it usually has nothing to do with the insides of the drive itself. In theroy, if you know how to work on surface mount stuff, I think there is a non-conductive freeze stuff you can spray on it to determine the bad part, and replace it, if you can find some specs of what it is

[stj]

Re: HDD issues.

i take it you have been over the soldering with a magnifier?

[Spork Schivago]

Re: HDD issues.

This cycling power issue by connecting it up inside still does not answer the question of if it is a hardware or software issue. Both issues can prevent it from performing on the USB port as both hardware and software are independent from the inside connector. I would say software routines on the two different ports are handled differently and I am sure the hardware is different. You could have more than one problem, that is because you tried it on the USB port, which has it own driver. When the driver sees something corrupt by either software or hardware then it performs a certain routine. All depending on who wrote that driver and how good of a program is depends if the system response makes sense or if it ends up in some un-recoverable loop. So with that being said once you connect to the SATA port then with a different software routine you could cure the power cycling routine and still have the original problem of a corrupt program, software or bad sectors on the disk.

Oh wow, that makes sense and you bring up a good point there. I just assumed that because I hooked it up to the SATA port in the PC, when the problem disappeared, it was a hardware issue with the enclosure. I could some of your ideas with a good drive. However, the problem might only occur with bad drives.

I know the drive has dropped down to UDMA/33 now. ddrescue is just about done.

[Spork Schivago]

Re: HDD issues.

The "slow fix" doesn't speed up the drive per se. It just stops the drive from bogging down while doing error correction.

As for SAT, that is irrelevant for eSATA. USB uses SCSI commands but eSATA and SATA use ATA commands. That's why SCSI-ATA Translation (SAT) is required for USB.

I gotcha, thanks. The stuff I saw a few years back wasn't about speed I don't think, come to think about it. I think it had something to do with prolonging the life of the blue's by disabling something. Man, wish my memory wasn't soo messed up. I got a bunch of memories that just kinda collide together. You know, two separate memories that are instead remembered as one.

[keeney123]

Re: HDD issues.

In post #16, I post how when I hooked up the hard drive to the SATA ports inside the desktop, powering it via my desktop's PSU, the drive stopped cycling power. This appears to of fixed the original issue and answered my question. There are bad sectors on the drive, ddrescue shows that. Maybe I didn't word it correctly in post #16 or you missed reading it somehow? This has been a pretty hot topic, lots of good posts and good information. Really appreciate everyone jumping in and giving me suggestions!

This cycling power issue by connecting it up inside still does not answer the question of if it is a hardware or software issue. Both issues can prevent it from performing on the USB port as both hardware and software are independent from the inside connector. I would say software routines on the two different ports are handled differently and I am sure the hardware is different. You could have more than one problem, that is because you tried it on the USB port, which has it own driver. When the driver sees something corrupt by either software or hardware then it performs a certain routine. All depending on who wrote that driver and how good of a program is depends if the system response makes sense or if it ends up in some un-recoverable loop. So with that being said once you connect to the SATA port then with a different software routine you could cure the power cycling routine and still have the original problem of a corrupt program, software or bad sectors on the disk.

[fzabkar]

Re: HDD issues.

The "slow fix" doesn't speed up the drive per se. It just stops the drive from bogging down while doing error correction.

As for SAT, that is irrelevant for eSATA. USB uses SCSI commands but eSATA and SATA use ATA commands. That's why SCSI-ATA Translation (SAT) is required for USB.

[Spork Schivago]

Re: HDD issues.

At this point everyone is guessing what is wrong. Somehow you must eliminate the possibility of the software causing the problem. So , you software people out there how would he eliminate the software as possibly causing the problem? Once you eliminate that possibility then I would tear into it and proceed with the other hardware suggestions.

In post #16, I post how when I hooked up the hard drive to the SATA ports inside the desktop, powering it via my desktop's PSU, the drive stopped cycling power. This appears to of fixed the original issue and answered my question. There are bad sectors on the drive, ddrescue shows that. Maybe I didn't word it correctly in post #16 or you missed reading it somehow? This has been a pretty hot topic, lots of good posts and good information. Really appreciate everyone jumping in and giving me suggestions!

[Spork Schivago]

Re: HDD issues.

when doing data recovery, u want to get the data off in the fastest and most glitch free method. so use a native sata connection.

doing data rescue via usb is more prone to glitches than a native sata connection as u are at the mercy of how well or sheety they designed the usb bridge/sata converter. also, if the connection is usb2, u slow down the recovery process. u want to get the data off as quickly as possible and not kill the drive during recovery.

the -n switch is for grabbing all the data off the good sectors as quickly as possible. then if the drive is still alive after the initial mass data grab, u can now try the recovery of the bad sectors.

this is especially important if the drive has tons of bad sectors in the beginning of the drive. for example, u dont want the program to stay stuck at the beginning 5gb of the drive trying to read bad sectors. what about the other 745gb of data? u want to prioritize all the good sectors which waaay outnumber the bad considering the capacity of drives these days.

the -n switch is also useful for performing extreme data recovery methods thereafter. an example of an extreme deep freeze hard drive data recovery method is to use -n for the first pass to grab all the data off the good sectors. then coat the hard drive in vaseline (to protect against condensation) and chuck it in the freezer overnight. then in the morning quickly plug in the drive and quickly try to recover only the unread sectors (which are all the bad sectors) while the drive is frozen at subzero temps.

sometimes bad sectors become readable again during subzero temps but become unreadable again when the drive heats up. so if the drive heats up and sectors become unreadable again just keep repeating the freezer trick until all the bad sectors are recovered and voila u have saved 100% of your data!

Thank you for the explanation. I try to avoid the freezer unless I absolutely have to. Had to do it with a floppy disk not too long ago and was able to successfully recover the data. So, once ddrescue finishes, how should I restart it to try and recover the bad sectors? Seems the longer this has been running, the error size reported by ddrescue has grown quite large. Last time I checked, over 238MB! However, almost finished. The partition was only 300GB (and I'm at around 280GB I believe). Kinda weird how this drive was partitioned. 300GB partition they where using, but where's the rest of the space?? If there's another partition, why can't I see it? 300MB + 600MB + 300GB + 2.9GB + 20GB != anything near 750GB!

Oh no! sdc5 is supposed to be around 400GB according to fdisk. That one must of failed as well and only copied 2.9GB. I really need to get this hibernation file and pagefile stuff recovered along with the empty data. It is kinda important. If I find what I think I'm going to, I'll let you guys know. It's nothing good, that's for sure.