I've got one box chugging away as my main server, serving all my local tarballs over nfs to my other boxes. I've also got another box, to eventually become my new server (needs a cable so I can test my tape drive with 2,6 kernels, and needs me to get my pure64 build working reliable), on which I make backups with rsync to an nfs mount.
Three weeks ago, a bzipped binutils tarball went bad [ build failed for other reasons, corrected my script, bzip2 reported corruption ]. On that occasion, I went back to a mirror, downloaded a fresh copy, and discovered that both my original file and all of my backups had a different md5sum from the new download (I think I was logged into the server, and mounting the backups over nfs from the new box). Tonight something similar happened with a gcc tarball (two desktop machines both reported it corrupt, after a build failed because of a thinko creating a symlink in my script). This time, I untarred a backup from the new machine onto one of the desktops without trouble, then restored the backup to the server and untarred that ok. Nothing untoward shows in the server logs, and I haven't noticed any other problems in the intervening 3 weeks. Anybody seen anything like this ? The drive continues to pass the SMART tests. I wonder about a failure in the memory, but last time I had memory fail on me the box suddenly started oopsing all over the place. Ken -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-chat FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
