Does the CopyFile function verify that the data reached its final destination successfully?

re: Does the CopyFile function verify that the data reached its final destination successfully?

Daniel — Tue, 25 Sep 2012 12:37:39 GMT

TeraCopy does create some checksum for each file it copies/moves.

re: Does the CopyFile function verify that the data reached its final destination successfully?

Pentium100 — Mon, 24 Sep 2012 19:03:32 GMT

There was a bug in Server 2003 that corrupted files written to shared folders (corrupts on receive). After that, I started using a program that verifies copied files.

re: Does the CopyFile function verify that the data reached its final destination successfully?

Danny — Fri, 21 Sep 2012 09:22:18 GMT

"If the call succeeds, am I guaranteed that the data was copied successfully?"</quote>

Answer : NO!! No one will ever guarantee that..ever. Why? Liability. Q.E.D.

re: Does the CopyFile function verify that the data reached its final destination successfully?

Timothy Byrd — Thu, 20 Sep 2012 21:40:58 GMT

At a previous company, I was responsible for putting the demo version of our product on our web site.

One day, we started getting complaints that the demo would not install.

My copy of the installer worked, but the copy I downloaded from our site was corrupted, so I re-uploaded.

After that, I downloaded again - the file was still corrupted, but differently.

I tried downloading the file several times in succession, and each copy was slightly different.

I called into hosting provider for support, and the kid who answered said he'd rebuilt the kernel on the server to make downloads more efficient...

re: Does the CopyFile function verify that the data reached its final destination successfully?

satan — Thu, 20 Sep 2012 19:43:49 GMT

SATA error are written to event viewer instead of propagated back in function calls.

re: Does the CopyFile function verify that the data reached its final destination successfully?

JamesJohnston — Thu, 20 Sep 2012 15:24:36 GMT

I once had a network card whose default network drivers bundled on the Windows 95 CD would randomly fail to verify checksums when the card was under load, and pass along corrupt data as if it was correct. That was fun to figure out. Fix was to install newer drivers from the vendor since the ones on the Win95 CD turned out to be junk. I even think there was an MSKB article about this but I can't remember.

re: Does the CopyFile function verify that the data reached its final destination successfully?

Crescens2k — Thu, 20 Sep 2012 13:48:18 GMT

@Gabe

Of course, if you look at the frequency of corrupt data you actually get then it makes you wonder if how much of an impact those numbers have in real life.

I can't remember getting a corrupt file that could be blamed on the network from the internet in the last year (the corrupt files that I have received could be accounted to something else, like a congested server truncating files or the owner the broadband hardware doing maintenance and that caused all of the users to have problems for a few minutes).

My LAN has had an even better track record, I have never had a corrupt file transfered over this network at all, and it has been active for years. So if this "one packet in 10 billion" ends up with you never getting an error on a LAN or only around one filea year from the internet being corrupt, then I don't see why you are making such a fuss.

re: Does the CopyFile function verify that the data reached its final destination successfully?

Gabe — Thu, 20 Sep 2012 13:02:52 GMT

Ian: They say: "After eliminating those errors that the checksum always catches, the data suggests that, on average, between one packet in 10 billion and one packet in a few millions will have an error that goes undetected."

They do NOT talk about there having to be an error in the first place. Even so, there are over 15 hops between me and this blog site (just to take an arbitrary example), at least one of which is a wireless link over unregulated frequencies (WiFi).

If errors were uniformly distributed, maybe you can assume they will be rare. However, that's not the case: some links are just more error-prone than others, whether it's a bad hardware, buggy software, or EMI. The more hops there are between two hosts, the more likely that one of them will be over one of those bad links, which drastically increases the odds of undetected errors.

re: Does the CopyFile function verify that the data reached its final destination successfully?

Matt — Thu, 20 Sep 2012 10:06:19 GMT

@Gabe:

The checksums in Network protocols are designed to detect common network bitpattern errors, they are not designed to detect all accidental damage caused on the network, and don't even try to prevent malicious damage caused on the network.

If you don't like TCP's checksum, you can always upgrade it using the TCP-Alternative-Checksum-Request TCP Option. Alternatively, you should probably be using SMB or IPSec for your traffic, both of which upgrade their checksums to cryptographic hashes and make your file unreadable and tamper-proofed from evil people sitting on your network.

But CopyFile won't do that for you. CopyFile just hands the bytes off to the filesystem driver, who will hand them off to the network driver. What the network driver chooses to do with those bytes is up to it.

re: Does the CopyFile function verify that the data reached its final destination successfully?

AC — Thu, 20 Sep 2012 09:46:32 GMT

I once tried copying several large files over a WLAN network (desktop and laptop in the same room!) simply by using explorer and shared folders. My reasoning was that as the wireless reception was near perfect and TCP/IP has the mentioned checksums, this should work without problems.

But after several tries, the files always came out corrupted. I'm still not sure why the TCP/IP checksumming didn't prevent this to happen almost predictably.