On 06/15/15 02:46, Bernd Schoeller wrote: > Hi - > > I have got an OpenBSD box, and I would like to create regular full > backups of that box to a Linux server at a different location. > > The main purpose of this backup is to be able to restore the OpenBSD box > on a severe hardware failure (HD corruption, fire, etc.). If possible, > the backup should be incremental as I am somewhat bandwidth constrained > between the two sites. > > There are a number of remote backup systems floating around > (rdiff-backup, rsnapshot, etc.) and of course there are in-house > solutions (dump/restore), though I don't know if these are interoperable.
My experience with third party differential backup systems wasn't that it had issues across OSs, but that it had issues across versions of the software. I consider that a complete failure. In my mind, here's the issue: how well do you know OpenBSD and your OpenBSD machines (same question for any other systems you are going to be backing up) * If you understand your systems, you probably know all you really need is your config and data files to rebuild your system. * If your task is to make it so a non-OpenBSD competent person can follow a script and magically restore generic unknown system, you want something that will restore precise and functional system. In the first case, an rsync-based backup is probably almost impossible to beat. Combine with the --link-dest option (google for it. the man page is accurate, but you won't probably understand the full implications of this. When you are grinning from ear-to-ear and saying "oh wow" over and over, you got it), you can have rotated backups with minimal BW and disk usage, AND files on the backup system are directly viewable and usable (and every backup after the first is incremental, and every backup directory is a "full"). I've used systems like this for over a decade now, and I can't over-state how powerful and useful they are BEYOND simple backup and restore, I keep finding new uses for this type of system. Downside: can't really do bare-metal restores, and when crossing OSs, I've had issues with ownerships and permissions. If you want a mindless "administrator" (i.e., management candidate or administrator-of-the-month) to be able to follow a set script and restore a machine regardless of of the purpose and operation of the machine, you probably are going to be stuck using dump over an ssh link. All the OS specific issues are managed on the machine being backed up, all that ends up on your Linux machine is a big file that can be pumped back over ssh to do a restore. Downside: those dump files are not very useful for anything other than restores. Nick.
