You are not logged in.
We want an automated and secure background update process that doesn't disturb the work from our staff. Here is my idea:
Prerequirements: a CoW filesysten like ZFS or BTRFS
1. Create a clone (ZFS)/snapshot (BTRFS) from the current system
2. Mount the clone/snapshot
3. Chroot into mount
4. Update and config highstate (Salt, Ansible, etc.)
5. Some checks for correctness
6. Exit chroot + unmount
7. If checks successful -> change grub config, etc. to boot next time to this clone/snapshot
8. Clean clones/snapshots (keep x versions)
9. Inform user to restart PC on occasion
-> If the update gives a problem, some grub submenu show old clones/snapshots you can boot from
-> Since the current system remains untouched, no firefox restart, etc. is required.
I tested it and it seems to work properly.
Do you see any problems that I missed? Or an already existing solution?
Offline
Step 5 is the critical one and should not be automated.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
(A works at time B) && (time C > time B ) ≠ (A works at time C)
Offline