So this is part 3 of my odyssee
I have found a solution for almost every redundancy task in my project, beside one.
Two machines run Proxy, DNS and an ftp-daemon. Proxy redundancy and DNS are straightforward. Both machines run simultaneously. Heartbeat/CARP or similar will monitor the services and switch back and forth, they run independent processes without the need of syncing. But the ftp is tricky.
As drbd does not seem to be available for neither OpenBSD nor FBSD, how about a RAID-1 over the network:
Host 1 is active, Host 2 passive. Host 1 runs its RAID-1 with a local partition and a mounted file-system (possible?) Host 2 provides the NFS share the mirror resides in.
Host 1 fails!
Host 2 detects the outage: It kills off every connection to the failed Host 1, mounts the mirrored image and continues operating as active.
Host 1 comes back up!
Host 2 detects Host 1 as alive, mounts Host 1's mirror image as part of its RAID-1 and the array should rebuild itself, hence syncs up the Host 1 mirror. Host 2 remains active.
If Host 2 fails when active, the whole procedure is done vice versa.
What do you think?