So I’m fixing to put my new testing server into revenue service, hosting my own personally-owned sites.
I’m then planning to wipe and reformat the losing server (mainly because I know it has immaculately-clean IP addresses), and move a client who’s currently on CentOS 7.9 and cPanel to Rocky Linux 8.6 and Virtualmin.
Before doing so, however, I’m asking for advice on backing up the whole server, in addition to the backups that Webmin and Virtualmin do. I figure it would be a good time to come up with something perfect, since I’m ahead of the curve time-wise.
Should I just go old-school and dd it? That’s what I’ve been doing, with the tarball stored remotely on Backblaze. I’ve never had to actually use it. Or would rsync (or anything else) be better? How about my hosting provider’s R1Soft solution? Has anyone used R1Soft? Is it worth the money?
I don’t mind paying for a good solution; but I don’t mind working for a good, free solution, either. My goal would be the quickest possible restores should the server crash and burn in a spectacular way. If that costs money, then that’s fine.
i don’t back up the base os…i use virt’s backup system to backup everything else(in the pro version). it simply just works. If you want to backup literally the entire server run it inside a vm and backup the vm on a regular basis…
Thanks. I already do the Virtualmin backups and copy them over to Backblaze using rclone, and they work fine in test restores (including one very database-dependent site that I restored from CentOS 7.9 to Rocky 8.6 yesterday as a test).
It did, however, take me several hours to do the updates and customizations I needed to get the system ready to receive the backups the most recent time I installed it. I’d like to reduce that to something measured in minutes.
The servers are actually VPS’s, and the DC does weekly images. That would be fine for my own sites because they don’t change very much except for the mail on one site, page additions and edits that I could just re-upload on most of the others, and a database that I could restore from hourly backups on another. But I would like something faster, easier, and more automated for servers hosting client sites.
The DC also offers R1Soft, which is supposed to create de-duped backups on a more-or-less continuous basis, with point-in-time restores and retention based on how much storage the customer buys. I haven’t tried it, but am open to doing so given some positive reviews.
I suppose I could also write a shell script that would take the system from a fresh OS and then do the OS updates, Virtualmin installation, and customizations, to automate the system recovery from fresh OS to being ready to retrieve and restore the Virtualmin backups from Backblaze. I could probably get the whole system recovered in under an hour or only slightly more that way.
I think I’ll play with that option on a local testing server. If I can automate the recovery from the point of installing the OS to being ready to receive and restore the backups in an hour or less, I’d be happy with that.
