> Pretty much everything Ed wrote is a good summation of the previous > discussion. To summarize the summary: LVM2 volumes are volatile. If you > are not very careful about this volatility then you could screw yourself > over and you won't know it. And even if you are so careful you can still > end up screwed.
How? > > There is one particular case of this that makes me worry: loss or > corruption of an incremental backup. Because you are backing up the block > device underneath you are backing up the raw blocks containing the file > system metadata. If you lose an incremental backup in the middle of your > cycle then attempting to restore subsequent incrementals will leave that > metadata in an inconsistent, possibly irrecoverable state. In a worst > case, loss of one incremental can result in the loss of almost an entire > backup cycle's worth of work. That can happen in a poorly designed system, sure. A file system "incremental" backup is merely an optimization. A full "image" of the source should be effectively backed up. By using the immediately previous backup as a source for the full backup, you can record the data for the full backup. At any point in time, you should be able to restore a "whole" in one operation. at any point in time, if you have a failed incremental, a full backup restores the state. > > --Rich P. > > _______________________________________________ > Discuss mailing list > [email protected] > http://lists.blu.org/mailman/listinfo/discuss > _______________________________________________ Discuss mailing list [email protected] http://lists.blu.org/mailman/listinfo/discuss
