On Mon, 2006-04-10 at 06:45, Timothy Murphy wrote: > > OK, but why should it hurt anything to have the destination > > file system have a copy of the running source system? Shouldn't > > anything necessary be reconstructed and anything else not > > matter? If you can't restore a working system from a tar > > backup a lot of people have a big problem. > > That's a very good question - in fact, more or less the same question. > You will have to exclude some directories, eg /proc, if you do this; > so the same question arises, which do you exclude? > > Actually, I didn't try tar on this occasion. > I used rsync simply because it is easier to exclude directories with this, > although I know it is possible with tar also. > Well, there was another reason for using rsync; > I thought some problem might arise during the copy, > and rsync would allow me to resume. > But is there actually any difference between rsync and tar > for this purpose? As long as you use the -a option with rsync it should do the same as tar - and you need -H for hardlinks and -S for sparse file handling, but omitting them shouldn't be fatal. Normally with either one I would use the --one-file-system option and repeat for each mounted file system I wanted (which may or may not be separate on the target copy). That automatically skips /proc and any odd CD or nfs mounts that might be active. But, I haven't done it with udev based systems. -- Les Mikesell lesmikesell@xxxxxxxxx
