Hi, and thanks to both jim and nightstrike for quick reply. If I wasn't clear enough, the files are on a Seagate FreeAgent external USB HD. The system is not over-clocked, and should not be too hot. As suspected, it seems as it has nothing to do with md5sum, but I reply with my findings, anyway.
It seems as writing to the disc is always successful, but reading the disc will cause apparently random errors, allthough there is no error message in dmesg. However, if I issue "modprobe -r ehci_hcd", the md5sum is equal every time. I guess this means that it is a HW/ehci_hcd-issue? What makes the whole thing so weird, is that I have another folder on the disk, with the same type of files, (DV-video from dvgrab), in which md5sum always produces the same outcome, i.e. no reading errors. I have ran fsck.ext3 -cc on the partition. Anyway, my problem is solved, as long as I use uhci_hcd, but I never realized that USB2 was so much faster than USB1... I will contact someone else if I get too tired of the low speed. 2008/3/15, Jim Meyering <[EMAIL PROTECTED]>: > > "Tarje Killingberg" <[EMAIL PROTECTED]> wrote: > > I am getting variable results from md5sum with the -c switch. I suppose > a > > bug in md5sum is unlikely, but maybe you have some ideas for what I can > do > > to find out what is going on. I have included some commands and their > output > > to describe the situation and some other information. > > > > # md5sum -c test06.avi.md5 ; md5sum -c test06.avi.md5 ; md5sum -c > > test06.avi.md5 ; md5sum -c test06.avi.md5 ; md5sum -c test06.avi.md5 ; > > md5sum -c test06.avi.md5 ; md5sum -c test06.avi.md5 ; md5sum -c > > test06.avi.md5 ; md5sum -c test06.avi.md5 > > test06.avi: FAILED > > md5sum: WARNING: 1 of 1 computed checksum did NOT match > ... > > > test06.avi: OK > > test06.avi: FAILED > > md5sum: WARNING: 1 of 1 computed checksum did NOT match > > test06.avi: OK > > test06.avi: OK > > test06.avi: FAILED > > md5sum: WARNING: 1 of 1 computed checksum did NOT match > > # cat test06.avi.md5 > > 002bf7b936439242c8b8c85a3d37a5de test06.avi > > > Ouch. > Suspect your hardware: memory, CPU, disk. > Could the system be too hot? > Is the CPU over-clocked? > Let memtest run overnight. > > If you have enough space on a memory-backed file system > (usually /dev/shm), copy the file to it first and repeat the experiment. > That should eliminate the possibility of disk read errors. > _______________________________________________ Bug-coreutils mailing list Bug-coreutils@gnu.org http://lists.gnu.org/mailman/listinfo/bug-coreutils
