[EMAIL PROTECTED] (Larry Jones) wrote: > Essentially *every* instance of repository corruption we've encountered > over the years has been caused by NFS interoperability bugs. I'd think > very carefully before putting a repository on an NFS mounted disk.
I tracked one case of repository corruption to a bad network switch that would corrupt packets on e.g., checkout or commit. The resulting file (working copy or ,v file) would have the proper length, but with a couple groups of four bytes moved around in such a way that basic networking checksums were unchanged. Needless to say, this didn't happen often, but when it did, it was a pain. Reproducing it took lots of iterations. NFS was not involved. This was all via ssh and local disks. _______________________________________________ Bug-cvs mailing list Bug-cvs@gnu.org http://lists.gnu.org/mailman/listinfo/bug-cvs
