Just Work

Today I experimented with using rdiff-backup to pull the data for the backup. I was curious to see if this method offered any advantages over pushing the data to the backup server. I was using a recipe so the setup wasn't difficult, but in the end I couldn't see any benefit and decided that pulling actually has some disadvantages.

The backup server will need root access on the system to be backed in order to preserve file permissions correctly. The ssh forced command in the authorized key makes this safer but allowing remote root login goes against my normal practice. The recipe didn't cover recovery at all even though the procedure actually becomes convoluted and counter intuitive. The simplest way is to use root on the system needing recovery, and login to the backup server as the user that runs the backup jobs, so the restore pulls files from the backup server. But, this is same as doing a recovery with the push backup method! The final nail was that I have a mix of Windows and Linux systems which require backup, so the pull method would require a ssh daemon on Windows. A daemon exists in Cygwin but I've never run any servers under Cygwin, so there are a few unknowns down that road.

I'm sure using rdiff-backup to pull data makes sense in some circumstances, just not for mine.