I learned the other day that if the backup server has a catastrophic disk failure, that there is no way to make backups for existing VMs without having an offsite copy of the data which existed on the backup server before it failed.
In most backup systems, if the backup server dies, you simply swap out the disk, and make fresh backups. As it was explained to me, with the OnApp backup system, the original parent templates and children of those parents must be preserved in order to restorer backups.
If this is true, the documentation does not mention anywhere that in order for OnApp backups to really work, you probably need a secondary backup server to protect against the first one's failure, or at the very minimum, a secondary storage (like a secondary disk on the backup machine) which backups can be copied to using a mechanism like rsync.
I feel the documentation here:
As well as the recommended requirements that you see here:
Both documents should clearly describe this potential catastrophe. This will also save network designers and hosting companies from having to redo their architecture multiple times, which is a costly process which could be avoided with proper documentation.
I am not sure that everyone is aware that if their backup server tanks, that their only option to take backups on VMs that existed before the crash, is to completely rebuild ALL VMs and all customizations from scratch, then to migrate web site data from the old VMs to the new ones. This pretty much defeats one of the primary goals of visualization.
A better fix would be to figure out a way to simply allow the backups to start on a fresh harddrive without requiring the original templates, but I am not sure if it is even possible.
Please sign in to leave a comment.