I have a server also running some systemd user services. Now one service runs a headless chromium browser, which is provided by Ubuntu snap on this system. It now runs under its own scope a la snap.chromium.chromium-....-.scope, not tied to the launching systemd user service scope anymore.
As so, all control over its run time is lost, and stopping / restarting the user service does not affect the spawned browser. Frequent restart on error condition quickly leaks chromium processes now, rendering the system unresponsive.
This is not specific to chromium I guess, but snaps in general I assume.
How can I force snaps to respect the cgroup environment it is spawned by?
