HDD Data Repeatedly Corrupted or Missing

[Pokerfish]

Some weeks ago, I replaced my 40 GB system-HDD, which was connected directly to the motherboard bus. In addition to the system drive, there was a 200 GB storage-only drive attached to an ATA-100 PCI card. It was divided into three equal 62 GB partitions ("D", "E", and "F").

I replaced the 40 GB drive with a new 200 GB drive which I attached to the same ribbon cable as the old 200 GB drive. I moved the old drive to the slave position, and installed the new drive in the master position. I set the jumpers on both drives to "cable select." (Previously, the old 200 GB drive was set to "master.") I installed WinXP SP2 and applications on the new drive, and everything was back to normal.

However, sometime later, I discovered that some of the contents of folders in the "F" partition were missing.
Then about two weeks ago, I opened the "D" partition, and although all the appropriate folders were there, their contents had been turned into various "gibberish" files and folders. None of these could be used or opened, and stranger still, the displayed file sizes were nonsensical (some folders were "100 GB" or other impossible sizes, given the partition size was only 62 GB). I tried copying the "D" partition at this point, but only got a "cannot read" error. Running chkdsk turned the drive contents into about 60 GBs worth of .chk files. These were virtually useless since the partition was fairly fragmented.

Then today, after reformatting the "D" partition, I later discovered nearly all the contents of the "F" partition had gone missing. I know it was all there, as I accessed the partition as recently as last night. After rebooting, Windows chkdsk ran during startup and deleted all the "orphans." Most of the contents of the "F" partition are now gone. So far the damage has been limited to the "D" and "F" (beginning and end) partitions. The "E" (middle) partition has thus far been unaffected.

Ok, obviously something bizarre is happening with the old 200 GB HDD. I ran the manufacturer's diagnostic utility after the "D" partition got corrupted, and the drive passed all tests (not requiring formatting). I also ran a virus scan and it turned up no viruses. The drive itself is almost three years old, and never had any problems until after I replaced the old 40 GB HDD. The BIOS, the PCI card startup dialogue, and the OS recognize the drive's full capacity (186 GB).

What could be going on, and what should I do to ensure no further loss of data?

Thanks for your advice!

 

[Pokerfish]

An update and a *bump*:

I've concluded that somehow the HDD's full 186 GB capacity is unuseable. The "D" and "F" (first and last) partitions seem to be competing for storage. Although Windows recognizes each partition as having a 62.1 GB capacity, if I fill one of them up, then add data to the other beyond a certain point, data from the full partition begins to disappear.

To reiterate, I was always able to use the drive's full capacity in the past. It is connected to an ATA 100 PCI card in order to overcome the 137 GB limitation, and for all intents and purposes its full "186" GB capacity is recognized by the system. Still, the disappearing data tells me something is amiss.

Any suggestions would be greatly appreciated.

 

[BrucesXP]

Hi

OK, it's a complicated issue and you've not posted a lot of raw materials. Here are a few general thoughts

1 With XPSP2 Assuming you are using a valid NTFS on both sets of drives you should have no OS problem with NTFS up to about 15TB if I recall correctly - you can actually push it further by increasing the cluster size.

2. The 137GB ATA limit could still be a problem if you are using an older MB or bios (it is slightly encouraging that the BIOS recognised the drive size correctly but it all sort of went wrong after that didn't it). The 137GB limit was overcome c2002 on new boards/bioses. The mess you describe is characteristic of the interface failing to address the two large drives correctly (look at the bottom of this page

http://www.dewassoc.com/kbase/hard_drives/hard_drive_size_barriers.htm

for the whole sorry tale.

3. I NEVER use cable select as results vary to say the least. I do not appreciate ANY variability regarding my drives

4. Without the full spec from your interface card I have no way of knowing if your new configuration is actually supported.

5. A look around the web will still find you a range of MB's that would take a 2-3 year old processor/memory/AGP combination but support large drives directly and correctly (e.g. still lots of Socket A options around at low prices even from the majors like Gigabyte) or you could use this as an excuse to spend a bit more and upgrade properly)

6. If the addressing is wrong formatting and write/readback testing might not work as expected.

7. My money is on the interface card being out of it's depth but if you need this amount of storage then get a MB designed to work it directly.

 

[Pokerfish]

BruceXP:

Thanks for the reply. Sorry if I wasn't clear in my first post, but I have had no problems in the past with the drive at issue. It is a 200GB Western Digital HDD and was previously the only drive connected to a Promise Ultra ATA 100 controller card (which came with the drive). The only thing that's changed is I've added a new 200GB Seagate HDD to the controller card (to replace the 40GB drive that used to be connected directly to the MOBO). I attached the new Seagate drive to the master position on the ribbon cable (where the Western Digital used to be), to use as my OS drive. I then moved the Western Digital to the slave position, because it's merely a storage drive.

So 2 things to keep in mind:
1) The 200GB Western Digital HDD worked perfectly before the drive-swapping.
2) Even now, the 200GB Seagate seems to be fine. The Promise boot-dialogue and WinXP recognize both drives at their converted 186GB capacity.
Therefore, there doesn't seem to be an issue with the system not overcoming the 137GB limitation.

All I know is that somehow, after doing the switcheroo, the old 200GB Western Digital began acting up. What I plan to do next is, once I'm able to back all the data up, I will use the Western Digital boot utilities to reformat/reinstall the drive, and see if that fixes things.

As to using "cable select," I suppose there's no reason to choose it if you know you're doing a master/slave setup. Just that some computer manufacturers (e.g. Dell) recommend using cable select--perhaps for inexperienced installers I suppose.

In any case, thanks for the suggestions. I'm definitely still open to more.

 

[BrucesXP]

Thanks for the info.

I am thinking (it is cold and wet in the UK so only slowly)
1) Just because the card can support one big drive does that provide any guarantee it can support two? (Spec?)
2) Just because the MB can be fooled into accepting one large disk full of extra address space does that mean it can accept twice that?

To explain both:

As I recall (Greater experts please chip in) the ATA controller you are using on the troubled channel (i.e. the IDE interface board's ATA channel) has changed from that in your old 200GB master drive to that in the new 200GB master drive - i.e. different type/firmware ? Also the old ATA controller probably relied to some extent on the associated interface board to deal with the MB over it's big size - the new drive maybe thinks it knows how to manage it on it's own?

You have no choice in this with the new drive on that channel- your C drive has to be a master (a flag given by jumper or automatically via cable select) and the flag "master" tells the drive having it "Use your embedded drive controller (aka IDE) to direct traffic to yourself and to the slave" Hence the usual folklaw of putting your newest/fastest drive as master on any channel - thus to provide the best/newest/fastest controller to the whole channel.

SO this should all work . . Yet it does not. And adding a new master on the channel has mucked up the (now)slave . . .The thing you have changed is the ATA controller for the ENTIRE channel but the big makers are fairly good at these by now and you'd really expect it to work OK especially if it passes it's maker's self tests. That then still leads me back to the ability of that card to broker understanding between your older MB and all this new generation controller electronics and address-space.

So much for logic - Other basics before heading off in rocket science would include
1) have you also tried sticking the interface card in a different slot and then updating/reloading any drivers (ideally with WHQL signed ones from microsoft)

2) Have you checked you are using the latest bios for the motherboard

3) Have you tried formatting and populating the drives (copy a load of install CD's to them for example) in another PC that can handle them directly and then running chkdsk in long and detailed mode when they are moved back to this one.


HTH

 

[garebo]

This is a bit over my head and i acknowledge that. However, there is one are that isnt. Whenever i have trouble with a hard drive i always use the hard drive mfgrs diagnostics program to write zeroes to the hard drive. I would say that well over half the time it solves my problems. Doing the zero write sets up a new mbr for one thing and this eliminates the possibility of a virus\trojan\worm at work, not that i think this is your problem. But the zero write seems to take care of a lot more problems than just setting up a new mbr, so i suggest you take that one more small step. Works for me a lot of times. And it has worked for lots of people posting hard drive problems here. Its one more way of making sure the hard drive is not the problem so that you can concentrate on other issues.
Just a suggestion.


Good advice + great people = tek-tips

 

[PLBinMTL]

Hi there.
I wish I could offer some help, but you'll have to settle for some moral support. I currently have the same problem with two of my drives. I just started running scandisk on the 2nd partition of my 1rst drive. I previously ran anti-viral checks, and reformated some of the partitions with partition Magic 8.0, and since both drives are Maxtors, I also ran the MAxBlast3 setup program and the PowerMax DOS diagnostic utility... no luck! Here are my machine specs; The Master HDD is a Maxtor DiamondMAx 120GB partitioned in 3. The Slave HDD is a Maxtox DiamondMAx 40GB partitioned in 2. I have a GigaByte mobo straddled with a P3 1GHz and 512 megs of RAM, running Win98SE. Both drives are less than 6 months old. I also remember that, about 8 months ago, when I had a dual-boot configured with Win98SE on one drive and WInXP Pro on the other, the HDD with XP started seeing files as "garbbled" and non-sensical and then, poof, the partition holding XP was no-longer visible on boot-up and I lost my dual-boot system. I since had that Maxtor 40GB drive replaced with the one that I have in my system now, the same one I already mentioned above!! Also I have, boxed in my closet, a Maxtor 20GB 5400 RPM with Win98SE pre-installed that I pull out whenever or if ever my system goes haywire, and let me tell you that it has never had this problem, which indicates to me that the problem, whatever it is, has probably nothing to do with my rig's configuration. My other machine cards, video,audio and ethernet all work fine, as do my two optical drives and floppy. I'm completely baffled by this issue. I guess you could say it's an on-going battle, and any help or insight would be appreciated. Good luck to Pokerfish and anybody else who may have this weird problem.
PLBinMTL

 

[Pokerfish]

BruceXP:

The controller card I'm using is the Promise Technology Ultra ATA100 PCI Controller Card. Different sources (e.g. the control panel on my computer and the box the HDD and card came in) give it slightly different names, but I believe it's the one at the bottom of this page, called "Ultra100 TX2":

http://www.promise.com/product/product_list_eng.asp?familyId=3

Unfortunately, the site doesn't seem very updated as most of the links to the products, including mine, are blank. Still, I'm fairly certain the card is designed to accomodate two IDE channels, capable of connecting two devices each (via standard ribbon cable). Moreover, the boot-dialogue for the card lists four device "slots," the first two of which, contain the two HDDs. The 3rd and 4th say "not connected" as I have nothing connected to the other IDE channel. The card was specifically included with the old 200GB HDD to overcome the 137GB limitation on older systems.

However, the two possibilities you point out are certainly valid. Indeed nothing is guaranteed, but without knowing more, I'd say that at least in theory, the PCI Controller Card is designed to accomodate up to four large-capacity HDDs.

As to your questions:
1) I have not yet tried relocating the PCI card to a different slot, but I don't think I want to either. It is currently in Slot 1, and recalling the initial installation, the card does not seem to work properly if installed elsewhere. (Three years ago: I originally installed it in Slot 3, went insane trying to figure out what was wrong, moved it to Slot 1, and everything worked like magic.)

2) My BIOS is current as can be.

3) I have not tried reformatting/repopulating in another PC that can connect the drive directly. I may consider doing this as a last resort, but before I do that, I'd like to reformat/repopulate on the current PC first, using the manufacturer's diagnostic utility, as I mentioned earlier, and as garebo suggests...


Garebo:

I definitely plan to use all the manufacturer's diagnostic utilities, including writing zeros. I will actually try to fully reinstall the drive using the utility too. The only thing keeping me from embarking on this is getting around to backing up all the data currently on the drive--nearly 200GB worth. I will update as to whether this ends up working or not. And I agree, it's definitely something to try (to ensure there's nothing wrong with the drive itself) before trying other things.


Thanks both of you for the suggestions! As always, I am open to more.

 

[Pokerfish]

PLBinMTL[u/]:

Haha, wow, you must have posted while I was composing my last post.

Your situation, at least your setup, sounds a little more complicated than mine. However, you seem to be using a much newer machine, and you don't have the lingering possibility that the 137GB limitation has something to do with your problems, since none of your drives exceeds that capacity.

The 40GB drive I replaced (the action that initiated my problems) was also a Maxtor DiamondMax. I concluded the drive was on its way out when I had difficulty reformatting it and reinstalling WinXP. It served me well though--about 5 years--with pretty heavy use during that time.

Anyway, as to your own problems:

I take it the garbled/disappearing data is still your issue, even after replacing the 40GB drive? Are you having this problem with both your 120 and new 40GB drives or just one of them? On which drive is Win98SE installed? When do you remember the problem first occurring--after setting up the dual boot system? After adding a new drive? Other major system change?

I notice you just started running scandisk. I'd caution you that if you have important data on that partition, you risk losing it or making it useless if scandisk converts it all to a bunch of ".chk" files--which happened to me earlier. If the data's important, you're best off not touching the drive/partition, not even with scandisk or chkdsk, and seeing if you can first copy the whole drive elsewhere. The more you make changes to the corrupted drive, the more difficult data recovery will be.

If worst comes to worst, I'd suggest backing up ALL your data from both drives. Then completely formatting and reinstalling them from scratch using the Maxtor utilities (including the write zeros function). Then reinstall the OS and other essential applications. Then populate the drives with data you already have backed up, use the system for a while, adding/deleting non-essential data to the drives, and see if the problem still occurs.

Good luck!

 

[PLBinMTL]

Hey Pokerfish.
Thanks for the advice, but I'm way ahead of you... If someone farts in an apartment in the building across the street, I'm burning 2 DVDs and 3 CDs of backups, just in case!! I decided years ago that...I WOULD MOURN NO MORE!!..for lost data. So, basically, I'm running a PC with one working (?) partition, my C:\ drive. Maybe I should specify my machine specs a little. My 120GB and 40GB drives are both on IDE-0, as Master and Slave respectively. If I want a dual-boot machine, I have the Master drive carry my preferred OS(usually Win98SE), and the Slave drive has the secondary OS(usually XP Pro). My IDE-1 channel has a Sony DVD-burner as Master and an LG CD-burner as Slave. All are jumper-configured drives. The HDDs are on a Ultra DMA 80-conductor/40-pin cable, as they should be, and the optical drives are on a regular 40-conductor/40-pin cable. I only have data on the C:\ partition right now and I'm currently having the Win98SE version of Scandisk go through the other four. If the problem persists, I'll just pop-in my pre-installed 20GB drive and get RMA numbers for the other drives, they're both still on warranty. I would like to resolve this issue without having to pull out and package the drives though. Anyway, on goes the struggle. I'll keep you posted. No pun intended.

Ciao, and good luck,
PLBinMTL

 

[garebo]

This is kind of an aside, fellas, but i have heard the word scandisk so many times i thought i would mention it.
First off, to each his own, i am only relating my experience and opinions, so no need to start a war on anyones part.
I dont use scandisk, havent for years, same goes with defrag! Honestly, I feel my drives are better off without them. As a newbie i did this and that and as i got better acquainted i developed by way of doing this, some right and some not so right,lol. Now i attribute the fact that i have had very very little hardware go bad on me to the fact that i have 5, yes, 5, UPS's and plug all pc parts into them. But i also credit the fact that i dont use norton systemworks, scandisk or defrag. I especially view scandisk as causing more h drive problems than it fixes. I stand firm in that conviction and nobody will change my mind as i believe i can prove it. I used to run those 3 progs religiously. When i stopped running them i stopped having hard drive problems, period! Again, i say scandisk is the worst of the worst. Now, for some reason that i dont know, chkdisk is not in that category. I use it and i find it helps a lot and doesnt cause problems that the others do.
Again, please, lets not start a war here. Just take my opinions for what they are. You might experiment down the road and use these progs a bit less and then think of what i said. On the other hand, i dont want to tell someone else what to do at all, just giving my opinion. Also, this isnt my thread. Perhaps i shouldnt have even posted, so lets not turn this into anything except one persons little rant, lol.


Good advice + great people = tek-tips

 

[kiddpete]

I've had the problem with the Promise Controller. You need to install an updated driver for it. The driver that Windows supplies only supports 137 gigs. If you figure out how to get Windows to go above that limit, you will trash your data. After the new driver is installed, everything will work correctly. Check the Promise web site for the latest driver.

Connecting the drive directly to the Mobo may also resolve the problem as long as the Mobo can handle the larger drive.

 

[Pokerfish]

UPDATE

I backed up all the data from the dysfunctional drive, ran Western Digital's utilties (a quick test, a full test, and a full format--writing zeros to the entire drive), then used the utilities to format and partion. This time I just partitioned the entire drive as one 186GB partition.

After that was done, I started populating it by copying the backed up data. Low and behold, as disk-usage approached the 137GB threshold, an error message interrupted the copying, saying something like "Cannot copy file xyz. The file or directory is damaged(?) or corrupted." After I clicked "OK" I could no longer open the partition. I got this error message:

D:\ is not accessible.
The file or directory is corrupted and unreadable.

So, what to do...


kiddpete

You took the words right out of my mouth. That's exactly what I was starting to suspect, but now that you've confirmed it, I'm hopeful again. I noticed that WinXP SP1 installed a driver for the Promise card. I was wondering if it would make a difference to install a driver from Promise.

The strange thing is that the new 200GB Seagate drive I installed as the boot-drive, which only has the WinXP driver, doesn't seem to have any problems. Disk usage on that drive is beyond the 137GB point (there are 23.3GB free out of 186GB), yet there is no missing data or corruption. I hope it isn't a disaster waiting to happen.

So you suggest I obtain the latest driver from Promise and update the current one? Will doing this on my boot-drive (the Seagate) solve the problem with my storage-drive (the Western Digital)? Also, if I can't find the drive from Promise (their site seems rather neglected), do you know if the driver that came with the drive and controller card (on a floppy disk labeled Version 1.0) will work?


Thanks all for the advice. I'll be back with whether or not the driver update works.

 

[Pokerfish]

OK, I should have tried updating the driver before my last post. I obtained the Promise Controller Card driver from Western Digital, here:

http://support.wdc.com/download/downloadxml.asp#8

It turns out it's identical to the version I have on floppy disk, that came with the drive and card.

After updating the driver, I reformatted the bad partition once more, then populated it almost to full capacity (only 267MB free!). No error messages. No corruption. No missing data, so far as I can tell.

Which leads me to my next inquiry: does anyone have a suggestion as to how I might verify the drive's integrity. How can I be reasonably sure data won't suddenly disappear on me a week from now? I've run the manufacturer's diagnostics several times already, with no errors reported, and I doubt scandisk will turn anything up. Any ideas?

If I don't encounter anymore problems--dare I say it?--the issue might just be... solved?!

Awaiting your wisdom...

 

[kiddpete]

The problem was that the driver could not handle the larger drive. As you saw, as soon as you topped 137gigs, it started writing data to the wrong place. With the updated driver, it's writing data to the correct place. If you can write beyond 137gigs successfully, which you say you have, you are done. There never was a hardware problem. It was a software problem. Just don't roll back the driver.