On Tue, Dec 20, 2011 at 02:01:48AM +0000, Ken Moffat wrote: > > Meanwhile, the md5sum of the version on my server has changed. So > it looks as if I've got bad RAM on the server, and it got accessed > sometime after I originally checked the md5. Sounds like very bad > news :-( > Weirder and weirder - ran memtest86 for 8 hours on the server, everything looks ok. Checked the md5 of the libbonobo tarball after rebooting, still corrupted.
The last time I had a problem with bad RAM on a server, my backups were corrupted (I use rsync to copy the files), so I checked the md5s for this file in the backups - all ok, renamed the bad file and copied a backup. Looking at the SMART data for the drive, it all seems fine. I'm now even more puzzled than I was before - seems to be random corruption of a file *on-disk*, which just happened to change the permissions of one file in a tarball. Maybe a lot more of the files have been damaged, maybe it's a single instance. ĸen -- 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
