I’ve been working on getting snapshots of our VMs. The qemu snapshots didn’t work so well, so I am now using lvm and it’s snapshotting capability along with a small backup script to make hourly backups (at least while I’m still trying out the new system). This is so that when the developers or admins on our VMs hork things up so bad the VMs won’t boot, I at least have *something* to roll back to.
Not being able to roll back my VM disk images has been very scary. It’s one of those cases where a more traditional, back-up-the-changed-files-in-important-directories system (which we have working in tandem with the lvm snapshot-and-backup method) just doesn’t cover it.
Now to see if my new script did everything it’s supposed to…