Hi,

I am new to cgroups, so bit struggling to get something done.

I have a systemd service which when started, runs a particular process as daemon in the background.
The daemon then forks several child processes.
I want to set an upper limit to memory and cpu of each child process, say 1G.
What I understand here is that the systemd service, daemon and child processes are also part of same parent cgroup and thus memory/cpu limit assigned to parent cgroup (systemd services cgroup) will get divided between child processes that are forked.

The requirement is that each child process should get 1G memory each and some amount of cpu and it should not be that each child process should compete with other child processes to get a share of resources assigned to parent cgroup.

If for instance I set MemoryLimit to 50G for parent cgroup, but there are 100 child processes that are forked, each child won't get 1G memory.
So basically I cant set a MemoryLimit on parent cgroup, since I am not sure how many child processes will get forked

Creating a cgroup for each child process with above limits, may not be a good idea I guess.

This has been duplicated under "Server". Removing from zero-post bump.

Would have been better to request a move of this by the mods.

I see this thread "over here" being marked [SOLVED] but without yet being solved "over there." (I posted a suggestion over there which is not a solution.)

I wish there was an option to delete a post or move it to another section.

As per post #2, moderators can move threads - next time you can use the "report" button to request this.