> > When I compared the two files with md5sum on each > > side, I got different checksums. > > I did so a split of the file to 5 GB parts and > compared: > > > > Solaris: > ... > > 199d15f5204c14a97d29f86b14cd154b cpartad > > > Linux: > ... > > 0509b68893216f894ee394c0ab212cd2 cpartad > > You could transfer the "cpartad" 5 GB part to the > other side, make sure that this time the 5 GB part > isn't corrupted, and "cmp .-l" the corrupted and > non-corrupted part. How many bytes are different?
I redownloaded the file cpartad and compared it with "cmp -l" with the old file: $ cmp -l cpartad cpartad.old 4962108197 271 371 $ bc ibase=10 obase=2 271 100001111 371 101110011 It seems that 5 bits are wrong of the 23GB file transfer. What scp should do is to do a full md5sum of the files on each side: 1. One time calculating the md5sum before sending the file 2. One time calculating the md5sum after have received the file on the other side, and comparing if the two md5s are equal. Then it would detect network errors, and it would be sure that the error would be on OS/Disk I/O level on Linux/3Ware RAID controller. -- This message posted from opensolaris.org _______________________________________________ opensolaris-discuss mailing list [email protected]
