Hi Michal, Thank you for your reply. I can grasp the point. Please let me ask more questions if I think of any. Besides, I will fix the name of slice in the picture soon.
BR-Matthew ________________________________ From: Michal Koutný Sent: Monday, June 30, 2025 1:45 PM To: Matthew Chae Cc: systemd-devel@lists.freedesktop.org; Christopher Wong; Fredrik Hugosson; Umut Tezduyar Lindskog Subject: Re: [systemd-devel] Question about the systemd slice hierarchy Hello. On Mon, Jun 30, 2025 at 07:45:53AM +0000, Matthew Chae <matthew.c...@axis.com> wrote: > I have some questions about "Example 1. Enabling and disabling controllers" > from the slice-related man page linked below. > https://www.freedesktop.org/software/systemd/man/latest/systemd.resource-control.html# > > First, under system.slice, there is a slice named b.slice. Is this possible? No, the name encodes the path in reality. > Shouldn't it be named system-b.slice instead? That would be indeed the proper name of the slice that is pictured there. You can fix it here [1] :-) > Second, under user.slice, there's a user@1000.service, and under that, > there are app.slice and session.slice. If you check user@.service definition, you find that it sets Delegate= -- this means that the service as seen from PID1 perspective was delegated the manipulation of its (service's) cgroup subtree. > Can a slice like app.slice exist under a service unit like user@1000.service? This app.slice is a slice unit of the _user_ instance of systemd. It's full system path would be: /user.slice/user-1000.slice/user@1000.service/app.slice (app.slice is a random internal subcgroup for PID1) > If that's possible, what would be a use case for this kind of structure? It's turtles all the way down. One use case is structuring of processes by user's desktop environment, some more info is in [2]. HTH, Michal [1] https://github.com/systemd/systemd/blob/main/man/systemd.resource-control.xml [2] https://systemd.io/DESKTOP_ENVIRONMENTS/
