Your cloud server runs on one of several host nodes. Its disk sits in the shared Ceph storage that all nodes can reach. This is why the cluster can start the server on another node when needed.
What happens in a failure
If the host node your server runs on fails, the cluster detects this and automatically restarts your server on another node. Your data is still there because it is stored three times in the distributed storage. For your system this is like an unexpected restart: running processes and unsaved data in memory are lost, after which the operating system boots normally.
So that your services run again by themselves after such a restart, set them up as system services that start automatically at boot (for example with systemctl enable).
Switching off: always via the customer area
High availability remembers whether your server should be running. If you shut it down inside the operating system, this counts as a failure - the cluster starts it again. If the server should stay off, use Power actions > Shut down in the customer area. A restart inside the system (reboot) or a reset work as usual.
High availability does not replace a backup
It protects against hardware failure, not against deleted files, faulty updates or attacks. Back up your data with snapshots and the external backup.