how to tell if your hard drive is failing + needs to be replaced

[dr bu]

yeah, creativity and the way it shows...
i installed 923s flush extension.
fine art and ingenuity at every startup 

[IIO]

reveal volume problems - not hardware problems.

yet you can be sure that your drive will run out if you have 2000+ crosslinked files which are no loger readable, because that is usually caused by dead sectors.

the discussion of dealin with harddisks physically damaged is not a mac discussion, because low level formatting tools almost all only exist exclusively as DOS program.

btw, i find it totally anoying to search for bad sectors ... it takes about one week for a 1.5 terbyte drive in a G4.

[OS923]

My experience with my drives is that Windows is faster to complain about bad sectors and the SMART is not as reliable as you may think.

I formatted my Freecom Mobile Drive XXS with OSX 10.4. It says everything fine. On my XP PC the SMART shows 1 reallocated sector. I format it with XP using the Freecom HD format program. The SMART shows thousands of reallocated sectors.

I formatted a WD2500JB with OSX 10.4. It says everything fine. I install this in a XP PC. It's whining all the time: bad blocks, bad clusters were replaced by Windows and so on. Everytime I run CHKDSK the list of pending sectors gets longer, but when I reformat the drive, the sectors are not reallocated, which means that they are false positives, correct me if I'm wrong.

Should I use a WD drive only with a Mac?

[nanopico]

Is it that windows complains faster/earlier or is it that OS X is correct?

[OS923]

I think that OSX is correct and that Windows complains too early.

[DieHard]

My experience with my drives is that Windows is faster to complain about bad sectors and the SMART is not as reliable as you may think

Well, unfortunately, I have to test up to 10 drives a day in 8 different testing servers, all day, everyday; and I can tell you that S.M.A.R.T. (Self-Monitoring, Analysis and Reporting Technology) is very accurate and a godsend. It is very wise to get the data off when a smart failure happens and replace the drive. Temperature, Rotational speed, read errors, and other data that SMART utilizes is cold hard data and statistics and if you continue to use a drive that SMART deems on it's way out (as many of my clients do) you can buy a week, a month, and yes maybe a year, but you are prolonging the inevitable and most likely suffering with performance issues during the wait for the drive to totally die.

Get the data off, buy a new HD, and give the magnets to the kids or grand kids to play with from the old drive... they are so cool.  The point is, bad sectors usually lead to more (remember the old MFM, RLL days when we low level format... that was different technology, any SATA or PATA drive that needs low level formatting is ready for the trash, it should NEVER be done, it is on the way out, error mapping for the new technology drives is done at the factory... once)

[OS923]

SMART is not reliable.

Each day Backblaze records the SMART stats that are reported by the 67,814 hard drives we have spinning in our Sacramento data center.
...
23.3% of failed drives showed no warning from the SMART stats we record.
...
SMART 189 – High Fly Writes ... Operational Drives: 16.4% The false positive percentage of operational drives having a greater than zero value may at first glance seem to render this stat meaningless.

[OS923]

Suppose that my HD is starting to fail, then how do you explain that my soft bad sectors are not 'promoted' to hard bad sectors when I reformat? (They are a small number.)

[DieHard]

Suppose that my HD is starting to fail, then how do you explain that my soft bad sectors are not 'promoted' to hard bad sectors when I reformat? (They are a small number.)

"Soft" bad sectors do not always lead to hard errors, since they can be generated from abnormal shutdown and other issues, so yes, if you are really into low level formatting to remap/tag physical errors, and you notice that the number of blocks are much lower than the original "soft bad sectors" reported... then great, you dodged a bullet. 

Unfortunately, I am paranoid seeing hard drives fail everyday, and having to do data recovery with ice packs; so I stick to the "any small anomaly, replace the drive" method and I ebay all the drives that are probably perfect, for those lurking for a good deal

[DieHard]

Server Bench 1 of 3 testing drives from 7:30am to 7:30 pm

[IIO]

reformatting - and the process of writing new data to it in general - will normally simply ignore bad sectors, and a bunch of bad sectors is absolutely normal.

however, if you have a disk which is 50% full and holds 1 millions files, and you notice that there are two dozen files which can no longer be read properly (but only slower as usual or no more at all), it is most likely that there are a few hundred thousand bad sectors on this disk.

in this case the reason of these bad sectors might be a head crash or a magnetisation problem, and this is a warning for a future growth of the number of bad sectors.

after a physical damage to a harddisk there is often shrapnel flying around in orbits, and this can easily cause further damage to the disk, for example when it is turned upside down.

such a disk should be thrown away, and not used for important data any more.

[mrhappy]

Server Bench 1 of 3 testing drives from 7:30am to 7:30 pm

Sounds like a LOT of testing... somebody must be 'mainlining' IB Prfofin!!!

[OS923]

I tried GSmartControl under Linux. The Help says that there can also be a problem with buggy firmware or the manufacturer didn't follow the standard. Then I saw that there's a newer firmware for my WD and XXS as well as for my motherboard. I'll try that first.

GSmartControl appears to exist also for XP. I'll try that too.
gsmartcontrol.sourceforge.io

[OS923]

I didn't update the firmware. I do this only if it's no more than starting an installer.

I installed W98 and it worked without error.

I installed XP, but it couldn't restart, because "NTLDR wasn't found", but when I booted from an USB stick then NTLDR was still there.

I created a partition in the last 8 GB of the HD. The rest was not partitioned. I installed XP and it works.

I installed Puran DiskFresh and let it refresh the entire volume, which was a bit weird, because you can continue to use the computer while it's working, on formatted as well as other areas. After 3 hours and 45 minutes it finished without error.

Is there something similar for OS 9 or OSX?

[IIO]

refreshing and/or low level formatting is usually a dos/win only job. the manufacturer of the HD should list somewhere which tools are available (it is mostly theirs which work best)

[IIO]

btw, to "refresh" on mac why not just copy files off the disk and then copy them back.

this way you can also make sure that file are all still readable, and "refresh" the whole disk by reformatting it, move files away from faulty sectors and eventually delete tons of old crap which you dont need anymore.

[OS923]

A copy on a Mac is not an exact copy. Aliases get in the mix. If you have an alias A to file B on volume C and you copy this to volume D then A on D will point to B on C instead of B on D. If you copy a keyring to another volume then it will create an alias to the keyring instead of a copy.

[OS923]

I tried Norton Utilities 16.
It says everything OK, and 0 K in bad sectors.
The SMART status shows 1 uncorrectable sector.

[IIO]

that´s right, the catalog and driver are things you cant copy, it will be created new. but you want to check that part for physical integrity before anything else. ^^

[OS923]

This may be one of those rare drives without spare sectors. And the bad sectors are not dealt with by reformatting, because writing zeroes will result in reading zeroes successfully, promoting the bad sectors to good sectors. That's how we get in a loop. It's better to write ones, which are then read as zeroes, which clearly reveals the bad sectors. That's why I tried the badblocks(8) test under Linux, saving the numbers of the bad blocks to a file. Then I used that file to format the disk as ext4. This incorporates the bad blocks into the file system. Then I installed Slacko Puppy 6.3.0.

Boot with Puppy CD.
Remove CD.
Type in terminal: badblocks -wvs -o /root/bad.txt /dev/sda
This took 28 hours and 16 minutes.
Mount the disk.
Do in GParted: create unformatted partition sda1
This actually creates an ext4 partition with bad sectors.
Type in terminal: mkfs.ext4 -l /root/bad.txt /dev/sda1
This creates an ext4 partition without bad sectors.
Insert CD.
Install Puppy.

If I want to install Windows, then I can type mkfs.ntfs -l /root/bad.txt /dev/sda1
Then Windows won't use the bad sectors.

I found 864 bad blocks this way, while SMART reported only 53.

This is also better than relying on XP's CHKDSK which doesn't have the /b option to rescan for bad blocks.