Investigate updater not restarting some qubes

[deeplow]

Description

The 4.2 updater integration does restart some qubes, but not all. This is undesirable behavior since we want to enforce all qubes to restart.

Steps to Reproduce

SDW 1.0.0rc2

Expected Behavior

All running qubes based on templates that were updated should have been restarted.

Actual Behavior

Only a hand-full of qubes actually restarted.

qubes which got restarted	qubes (still) running after updater finished with pending updates

Comments

• when I encountered this I confirmed that sd-{small,large} did in fact have updates pending (via apt list --upgradable)
• If this ends up being an upstream bug, for the time being we should just force-restart all workstation-based app qubes.

[deeplow]

qubes which got restarted

From the looks of this, the qubes which got restarted (sys-whonix and sd-whonix) were either because they were whonix-based (I doubt it) or because they were sys-* (i.e. provides_network=True). Let's investigate...

[deeplow]

At least on the GUI updater only the sys-* qubes are selected by default:

If this is anything to go by, my bet is that the default behavior on the CLI (qubes-vm-update) is similar.

[deeplow]

Turns out this change in behavior was a consequence of a bug we opened. In short, --restart now is equivalent to --apply-to-sys (see diff) and there is a new flag which should fix it. This flag is --apply-to-all. I'll open up a PR for this.