Hello:
In former participations in this and other forums, I was advised to scan my hard drives in DOS, not in Windows, in order to avoid bad influences (software running and using CPU).
Before, I was using HDDScan in Windows, and now, following this advice, I am using MHDD in DOS.
But there is one problem with MHDD: it detects slow sectors, but it doesn't tell you the delay in milliseconds of each one of these slow sectors. Furthermore, it doesn't tell you the adress of each one of these sectors.
MHDD saves a log file but doesn't include this info either. It only tells you how many sectors are in each delay range.

HDDScan does show the delays of the slow sectors, and the addresses, but MHDD does not.

I am not an expert using MHDD. Maybe I am missing something? Is there any way to get this information?

Also, I have tried Victoria for DOS, but the manual is in Russian language, and I can't read Russian.

With MHDD there should be a legend on the right hand side of the screen

do you mean the legend in this screen?



yes, but it doesn't tell you for each slow sector, what is the sector's address and the delay for this particular sector.
For example, if my hard disk has 8 green sectors, I would like to know the addresses of these sectors, and the exact delay of each one.

Hello,

In this mode, the erase delays, the mhdd can't measure the delay, because immediately starts to write if the delay is too high.

The measuring of each sectors is unimportant, because the delay can be different, depeding on the error recovery methode wich the drive used to correct the data.
This depeds on more things, like temperature, the drive orientation (upside down) and some like this...
Additionally, if the sector shows error during write, it can be internally remapped to an another, wich can be in different condition, as like the source.
So, the "sector nem:delay" table is useless.

The only rule wich you can follow:
If your drive have too many green+ sectors, throw away the drive, because this will be failed soon!
The drives are really cheap now.
Don't use these drives on important places!

Regards,
Janos

I got this screen capture from google. Its not my case.
In my case, its a simple read scan, without "erase delays".

ok, I understand that the delay in milliseconds depends of many factors, but still, I think it is significant to keep track of it. For example, if you get a slow sector with 250ms, but you know that 6 months ago this same sector was close to 250ms too, then nothing has changed, and you must not worry.

I think it is very useful to know the address of the slow sectors and the delay of them in milliseconds, just like Windows program HDDScan provides.

The problem is that HDDScan runs in Windows, and I prefer a DOS program.

This is not true, because the slowness depends first on the error category, and on the contents too!
I mean if you fill up the hdd with FF, 55, AA, 00 another sectors will be slow.

Regards,
Janos

Additionally, if the content doesn't change, the error type not change, the surface and the head condition can be changed too with the time....

oh, so the delay of a sector depends on the contents of that sector?
For example, if you write 000000000 to that sector, or 111111111 to that sector, you get different delay? very interesting, but still, I would like to know the delay and adress of the slows sectors. I normally fill my big drives with data (video or music) and then I let them stay without content modifications for years. So I think the slow sectors will always be the same (more or less) and I will be able to keep track of the evolution.

>>>
Additionally, if the content doesn't change, the error type not change, the surface and the head condition can be changed too with the time....
<<<
yes, precisely, I want to keep track of things like the evolution of the surface and head condition, that is, the physical condition of the drive.

Yes, because the delays comes from the drive's error correcting methodes.
Like ECC calculating, reading-retry in the next rotation, trying to seek off and back (reposition), trying to read a little off-track, etc...
This is done if the readed content have at least 1 bit difference!
But in this way, if the address marks, sector IDs, servos are not damaged, the delays depends only in content.



Not true, because the bit-difference can be come from the superparamagnetic effect as well, wich can be happened when the drive is offline...



Trust me, instead of drives.
I have seen a lot of times, one really good condition drive is left for some year, and after the owner turned up, the head is crashed unexpectedly.
This is the most strangest thing wich can happened. (caused by the surface's coating layer)

For drive condition tracking, there is the SMART table inside the drive.
This is the best way to track the rarely used drives, but yes, you need to read the entire drive on every power-up, for check the condition again.
If you want to store big data wich not changing, the best what you can do is build up one software raid for storing.
I prefer to use Linux, because for example in RAID4/5/6 if one sector is unreadable, the code is automatically corrected by writing again from the redundancy calculated content.

Additionally, You can better track a 365 day online drive than the rarely used!

Regards,
Janos

SMART is not enough. if, for example, the number of green sectors grows, SMART won't warn you. Only when a sector is unreadable it will warn you.

I don't want RAID because the controller (hw or sw) can have errors managing the raid and mess up the whole raid. Also, data recovery is more complicated in a raid. Also, it won't protect you from software errors. Only hw errors.

So, again, I want to scan the disk for slow sectors, and have the addresses of the sectors and the delay in milliseconds. In a DOS program. Is this possible?
With MHDD, Victoria, or any other?

I don't know about any program like this, because nobody write this, because this function is useless.
I can't help you, sorry.

By the way, i have thinked about this again!

If You only want to track the delays changes, You don't need really the log, it is enoug the overall table, wich on the picture!
The numbers after the green, brown, red is important for You!
If these values increased, the condition gets more worse.

Maybe this helps a little....

Janos

ok, I don't think it is useless to know the location of the delays.
If, as you say, the delays are so variable that they are meaningless, then maybe it is useless the whole concept of scanning the drive for slow sectors. That is, MHDD, Victoria, HDDScan... all of them useless...??


>>>>
By the way, i have thinked about this again!

If You only want to track the delays changes, You don't need really the log, it is enoug the overall table, wich on the picture!
The numbers after the green, brown, red is important for You!
If these values increased, the condition gets more worse.

Maybe this helps a little....

Janos
<<<<

Of course, you can do as you say. In fact, this is the main use for these programs!!
Janos, I feel that you don't know much about these programs...

The overall table gives you a hint of the overall health of the disk, indeed. But you can't focus in a particular area of the disk, bacause you don't know where the slow sectors are located (unless you videotape the scan as it unfolds, LOL).

I am sorry, but for me, it seems to You have missed one point:
These programs not for end users, these programs for test/repair/recovery services.
These are a little bit different purpose...
Thanks to the delays monitoring (and much more of course), we can detect one bad / weak head or we can figure out the problem is based on head or a part of the surface (like scratch or media problem).
This is why these programs exists.



We are reached a really big theme at this point, but i will try to brief for you:

When the drives are manufactured, the parts will have some values wich needs to fit into some limits.
(skipping the quality classification)
The media will have some factory defects (more class) wich will goes into the P-list or reserved zone for hide them.
But some type of media error are still under the limit and will stay in the user area!
These sectors are what You hunting. (not exactly but near)
The error will be appears when the matching content wrote to these sectors.
If You want to avoid this issue, You should buy top-class drives, like SCSI/SAS.
(Of course You can't avoid totally, but You can get more closer what you want...)

Additionally, when the drive write the data wich generates the error why not remapped?
This is the key question!
Because the readback-test passed!
But the lean bit will lost its value with the time or when the head write another bits next to this one...

Regards,
Janos

ok, I think I understand what you say: some sectors are such that, when a matching content is written, then the sector will be slow, because an error bit appears, but still the whole sector can be corrected. Other times, the content is not matching, and the error won't appear, and this sector will be totally normal.

Then, this sector will appear sometimes as slow, and other times as normal.

But still, it would be good to keep an eye on this sector, I mean, write down his address, and write down the delays too (sometimes appear, sometimes not) to see if there is a normal behaviour, or the delays are bigger and bigger with time.

Ok, thanks for your help, Janos, you answered my question: there is no DOS program with this property, so I think I will have to use HDDScan in Windows, additionally to MHDD in DOS. More work for me.

Janos, do you know if there is any program that is able to fill the disk with different patterns of bits?
For example, you fill it with zeroes 000000000000, scan, and then you fill with 11111111111, and scan again. And then with 01010101010101.... etc.

this could be useful, isn't it? because delays depends on contents as you well say...

I am affraid, this is a little bit more complicated.
If You see one green sector, this means this sector have some corrupted bits, not only one!
Additionally, you can't be sure about the delay will increased with time or will be the same, because if one critical bit will loose its value (like one on the ECC data) the sector will be UNCORRECTABLE immediately, and You can't read 512 Byte caused by this issue...

By the way, what you are planning, is not 100% solution at all!

Regards,
Janos

Yes, the "badblock" named program under Linux exactly do the same.
Write the content what you want, than read back and shows if some sector gets unreadable (including ECC checking).
I use this prog for testing, but the most useful part is if you read the smart infos for compare between steps!
(need to parameter like -vvw -p if i remember right)

You can give the content in decimal or hexadecimal format.

Janos

oh i forget to add this to my previous posts:

The delay (=error type and correcting time) depends on more things, and not just the content of the actual sector, but on the contents of the next sectors as well!
In this way, if you do the write/read test from 0 to 255 you are not finished your job at all!
(The raw bits, what the drive write to the platter is not equal with the content, because the drive uses pseudo-random changes for break up the continuity for better position tracking of the head.)

Janos

Yes, I know badblocks for Linux, but I still don't have any Linux computer.
Some months ago, I spent a bunch of hours googling like crazy for something like badblocks, but for Windows, and I couldn't find anything.

Badblocks is indeed a very interesting program to torture your hard disk and seek for weak sectors. When I have time, I will prepare a Linux computer




Oh, I see matters are really sophisticated in hard disks, and you, Janos are very well informed. I am sure you will be a good option as data recovery expert.

thanks for your help.