This is a discussion on CONFIG_FAIR_CGROUP_SCHED (cgroups) and subprocesses (fork...) - Linux ; Hi all, how does CONFIG_FAIR_CGROUP_SCHED (from cgroup feature in 2.6.24) behave in case a process bound to a group spawns other processes (via fork or similar)? Are they bound to the same group or not? If they are not bound ...
Hi all,
how does CONFIG_FAIR_CGROUP_SCHED (from cgroup feature in 2.6.24) behave
in case a process bound to a group spawns other processes (via fork or
similar)? Are they bound to the same group or not?
If they are not bound to the same group this mechanism is pretty
useless, or at least it does not work like a reliable jail-like system.
Consider that fork is heavily used in lots of apps, in particular server
apps.
Thanks in advance
linuxnewbie1234 wrote:
> Hi all,
> how does CONFIG_FAIR_CGROUP_SCHED (from cgroup feature in 2.6.24) behave
> in case a process bound to a group spawns other processes (via fork or
> similar)? Are they bound to the same group or not?
By default yes. However, with suitable permissions it's possible to
manually move tasks between groups. (This is done via a
pseudofilesystem, so all the normal filesystem ownership/permissions can
be used.)
Chris
