>>>>> "Edward" == Edward Ned Harvey (lopser) <lop...@nedharvey.com> writes:
>> From: tech-boun...@lists.lopsa.org [mailto:tech-boun...@lists.lopsa.org] >> On Behalf Of Edward Ned Harvey (lopser) >> >> Just before midnight, I destroyed the machine. And then I started doing the >> restore. Guess what time the backup script runs? Midnight. Guess what its >> behavior is? It notices the filesystem is completely different, so it sends a >> new "Full" and clobbers the backup destination. Hooray! :-) Edward> BTW, every mode of backup is prone to failure, so I generally Edward> like to use more than one type of backup. In this case, I had Edward> the machine snapshots, and I also periodically export the Edward> config (it's a firewall VM). So I was able to rebuild the Edward> machine from scratch and then restore the config. I think the root cause of your failure here is that you assume ONE full backup is all you need. Instead you should be keeping multiple full backups instead. Edward> So what kind of backup script clobbers the backup destination? Edward> It's my own script, and here's the algorithm: Edward> foreach filesystem as "fs" on source: Edward> if fs exists on dest: Edward> if source fs and dest fs have a snapshot in common: Edward> send incremental source fs to dest fs Edward> else: Edward> rename dest "fs" to "fs_to_destroy" Edward> send full source fs to dest fs Edward> if send completed successfully: Edward> destroy fs_to_destroy from dest Edward> else: Edward> // put it back the way it was Edward> rename dest "fs_to_destroy" to "fs" Edward> else: Edward> send full source fs to dest fs Here is should be: send a NEW full source FS to dest FS, leaving other Full(s) alone. _______________________________________________ Tech mailing list Tech@lists.lopsa.org https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
