Henrik Holst wrote: > Zenon Panoussis wrote: >> Alasdair G Kergon wrote: >>>> [root@njoerd lsat]# md5sum 006069_15meROT.tif >>>> 0ce8093accb771c66a0154b4d55fd90b 006069_15meROT.tif >>>> [root@njoerd lsat]# md5sum 006069_15meROT.tif >>>> fc2caf002b693c1283a10536fa976358 006069_15meROT.tif >>> Compare two files carefully and analyse which bytes are actually >>> corrupt. >> I cannot very well compare a file with itself, can I? That's what >> I'm looking at above, one single file returning different checksums >> every time. That makes rsync and the contents of the file on the >> other computer completely irrelevant. If we forget rsync and assume >> that the file was created on this machine, the problem is still there. > md5sum tool computes the md5 sum by reading the entire file. Copy the > same file to /root/fileA and then to /root/fileB. That should create two > different files. Then compare files in /root. Perhaps I'm missing something, but I really don't understand the point of comparing files. In /root/ the md5sum is always the same, so there is no corruption and nothing to look for. In lsat/ the md5sum keeps changing. So, if I do cp /root/fileA /root/fileB diff /root/fileA /root/fileB there will be no difference. On the other hand, if I do cp lsat/file /root/fileA cp lsat/file /root/fileB diff /root/fileA /root/fileB the files will be different, but that will only tell me that lsat/file was read wrong at least once and perhaps twice during copying. I know that already, that is the problem I'm reporting, not something I need to find out. Finally, if I would do cp lsat/file lsat/fileA and compare lsat/file to lsat/fileA byte by byte, I would find that bytes X, Y and Z differ. How does that make me any wiser? I don't care one bit *which* bytes are wrong, all I care about is to know *what* makes them wrong. Theoretically, it could be the RAM, it could be the CPU, it could be the hard drive, it could the kernel, it could be dm-crypt. I start eliminating: /root runs under the same RAM, CPU and kernel as lsat, so those can't be causing the error, or else I'd be seeing it in /root too. That leaves me with the hard drive and dm-crypt. fsck and badblocks say the hard drive is good. I run md5sum on this file and then on another file and yet another, and it fails almost every time on every file. If I had a bad sector or two, I'd be seing the error on some files, but not on all of them. So I rule out the hard drive (which BTW is just a month old) and that leaves me with dm-crypt. Alasdair, would you care explaining what I'm missing and what the point of comparing files is? Z --------------------------------------------------------------------- dm-crypt mailing list - http://www.saout.de/misc/dm-crypt/ To unsubscribe, e-mail: dm-crypt-unsubscribe@xxxxxxxx For additional commands, e-mail: dm-crypt-help@xxxxxxxx
