You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Initial support for cgroupv2 unified resources was merged in #2584.
With that in place, one can set, say, linux.resources.unified.pids.max = 42 which will make runc to write 42 to pids.max. The problem is, systemd TasksMax will still be unset/unlimited.
The issue can be solved by adding a map of some known "unified" parameters
to systemd properties.
Alternatively, we can forget about systemd properties and implement #2602 instead (which AFAIK is what crun does).
The text was updated successfully, but these errors were encountered:
OK so we have discussed it with @mrunalp and @giuseppe and looks like we can proceed with the plan outlined in #2603.
The biggest concern about switching systemd drivers to create a sub-cgroup is that software that uses libcontainer (kubelet etc.) might break. We could do it as an option but I don't think there will be any benefit.
Initial support for cgroupv2 unified resources was merged in #2584.
With that in place, one can set, say,
linux.resources.unified.pids.max = 42
which will make runc to write42
topids.max
. The problem is, systemdTasksMax
will still be unset/unlimited.The issue can be solved by adding a map of some known "unified" parameters
to systemd properties.
Alternatively, we can forget about systemd properties and implement #2602 instead (which AFAIK is what crun does).
The text was updated successfully, but these errors were encountered: