DATA ERROR When Copying Files

[Harves]

Having a problem when copying my personal files and documents to a backup drive. After it has been copying for a few minutes it brings up a dialogue box with "Data Error (cyclic redundancy check)" and I can't continue copying files. Have never seen it before, can someone tell me what it means, and how do I prevent this from happening again?

[Tannin]

This is nasty. There are several common causes, but probably the single most common is failing media on one of the drives, most often the source drive.

Get a working backup now!

[Harves]

Just copying the important files (one at a time) that I know have changed since my last back up. Fortunately there are only a few. I will back up to my external drive as well.

My HDD is partitioned C,D,E & F drives

C = Programs
D = Files
E = Edited Images
F = RAW Images

It is the D drive that is giving me grief... can it be repaired via reformatting, or is it slowly dying?

[Tannin]

In more detail, a CRC is for our purposes essentially the same thing as a checksum or a parity check. The basic idea is simple.

When a device stores some data, a summary statistic is calculated according to so pre-set method. That statistic is also stored. When the data is read back, the same pre-set method is used to calculate the summary statistic again, then the two summary numbers are compared. If they are not the same, there is a problem somewhere!

Let's work an example. Let's say I'm sending you a message by morse code, and the radio reception is bad. So we need a way to check that the message you get is indeed the same as the message I sent. So, at the end of the message, I count up how many letters are in it, and send you that number. You do the same: count how many letters there were in the message you got, and confirm that you have the same number. If you have a different number, you tell me to send the message again, and we check again. If the numbers are the same, chances are the message is OK.

We could improve on this by using a more sophisticated error check: for example, we could count up the value of the letters (using A = 1, B = 2, and so on), and send that total. Then we don't only pick up on missing or added letters, we also pick up on errors where we have the right number of letters but one of them was wrong (I sent a P, but you received a K, for example). We can improve on that method too, and in fact there is a whole branch of mathematics devoted to working out the best possible methods for this type of task.

Computing and telecommunications uses this sort of error-checking method all the time: modems do it, hard drives do it, DVD drives do it, flash cards do it, mobile phones do it - in fact just about everything that communicates does it. (Note that data storage is, in this context, a sort of communication: the system sends the hard drive a message saying "remember this stuff" and then, later on, it sends it a message saying "send me that stuff I told you to remember".)

Now, to practicalities.

CRC errors can be caused by a lot of different problems, but the most common ones are:

• Media problems. The surface of the hard drive has started to develop faults, so the data cannot be read reliably. Cure: get the data off as best you can, then replace the drive.
• Drive electronic problems. The read/write electronics of the drive are failing. Cure: same as above.
• Cable problems. The data cable is failing. Cure: replace the cable. A replacement data cable costs less than $5, so this is always the first thing you should do, even when you suspect that the fault lies elsewhere.
• RAM problems. The data read from and written to the drive is stored in RAM. RAM is also used to calculate and store the checksums. One of the first signs of marginally faulty RAM is CRC errors on data access. If you have recently upgraded or changed RAM in your system, this is one of the first things to look for. The actual cause can be the RAM itself, the RAM control circuitry (which is part of the main board), or less frequently, the power supply. Test by replacing the power supply with a known good unit, swapping out the RAM, and checking that the system functions normally using a spare hard drive (put yours to one side while you test), with a fresh operating system install. Use a known good optical drive to do the install, and fresh cables. Then replace the suspect components, one by one, until you isolate the problem. (Yes: this can involve a lot of work!)
• Power supply problems. PSUs are the single most common item to fail in PCs, but usually give different symptoms. CRC errors are usually a sign of some other problem.

There are other causes, but drive, cable and RAM problems are the most common. About half the time, maybe a bit more, it's the hard drive.

If in doubt, take the system to a good general computer technician. (Not any bloody tech, a good one.) You may also wish to consider sending it to a specialist data recovery expert - but be warned, this will cost you a small fortune: can run into four figures without a lot of trouble.

PS: just saw your second post. The above is a general-purpose respone to the first one which, seeing as I've typed it already, I'll post anyway.

[Kym]

What Tony said !!
Just because you have partitioned your disk does not mean you have separate disks.
Borrow another external drive if you can and re-save your important stuff (call SpaceJunk?).
Belt and braces applies now - do not overwrite any existing backups!
Don't panic. Take things carefully and slowly.
The sad part is that you will burn some hours on this - just be patient.

[Tannin]

It sounds like media problems, Harves, but it's too early to say that for sure. One hing we can say safely is get your data off before you do anything else.

The simplest way to check the drive without special tools is simply to format it (which of course erases all the data - so be sure you have a good backup first). If you feel certain that you can do this without risking any data loss, simply do a long format of the partition from the command line. If it reports any bad sectors, even one, then you need to test more fully, or else just replace the drive. If the format completes without error, then it is probably OK, but in your shoes I wouldn't trust it until I'd done some more comprehensive testing.

Bear it in mind that a 500GB replacement drive costs $130 brand new. What are your pictures worth?

Also remember that progressive deterioration of a drive almost never just stops: once it starts, it gets worse and worse, and it is not unusual for the entire drive to suddenly stop with no further warning. Again, get that data off it!

[Tannin]

As Kym says, working slowly and carefully is a very good idea. His advice is, as always, excellent. You'll be fine, just take it step by step.

[Harves]

Thanks Tony & Kym, a great explanation and information on the problem. My computer has two 750Gb HDD's, Drive one as I described, drive two is purely for backing up drive one, which is usually done on a weekly basis. Then at the end of every month I backup the second drive to a external HDD. So fortunately loss of data is not an issue.

I am also noticing that accessing the suspect drive, it is performing very slow when looking up files and folders.

[Tannin]

That's another classic sign of impending failure, Harves. The reason will be that it's doing a lot of retries. (Drives try to read the data several times before they give up. This takes time, so much time that even a human can see the difference.)

You never get a final answer before all the evidence is in, but my money is now firmly on media problems. The drive will be under warranty, but unless you have a very good local supplier, you will have to wait weeks or months to get the replacement, and even then, unless it is a Samsung or possibly a Hitachi unit, it will be a factory refurb with poor to doubtful long-term reliability. So you might as well buy a replacement unit now. When the replacement drives comes in, you can use it as a second backup. Probably best not to trust it as first-line storage.

[Harves]

Fortunately it's a Samsung, only a few months old......... 3 year warranty on the unit, so that won't be a problem.

[Tannin]

Ouch! OK, you'll get good warranty service (at least I have always done so) but that's a nasty surprise. Samsungs genrally have an incredibly low failure rate - so much so that I now refuse to supply any other brand in the systems I build. We see maybe two fail a year, which when you consider the number we sell - enough to keep two of us in bread, milk, and lenses - is quite astonishing. Either you haver drawn a very short straw, or else your retailer isn't handling them properly.

[Harves]

UPDATE:- Downloaded a HDD diagnostic program from Samsung website, did a surface scan which found lots of bad sector errors. So it looks like it's faulty.

Performed a scan on my second HDD as well, it came out with no errors, goody

The problem I'm having now is that my computer cannot find my #2 drive in windows. It's registered in DOS-Setup, but windows can't find it for some reason. How do I go about getting it back into my windows directory?

[Tannin]

I'm not entirely sure I understand your question properly, Harves. One possibility is that it contains only what's known as an extended partition, not a primary partition, which is what Windows needs if it's going to start up from there. This is easy to fix, but it's very easy to hose any data on it if you make even a small mistake, so don't mess about with partitioning unless you have some experience with doing that and understand the principles. (Or unless there isn't any data to worry about, of course.)

[Harves]

Tony, I understand partitions, been using partitioned drives for years. I scanned my "Slave" drive, now windows can't see it via "My Computer" or "Windows Explorer" for some reason.

Anyhow..... I'll be off line for a while, sending HDD away now.

[ricktas]

See you on the other side Harves.!

[Harves]

Fortunately I can have a sneak peak every now and then from my work machine. So don't say anything naughty about me

[SpaceJunk]

waiting for a RA number Harves, first time I have had a Samsung drive go bad, will be interesting to see how quick the turnaround is