🔹 How are processes grouped into cgroups?

In Linux, processes are grouped into cgroups by:

1️⃣ Creating a cgroup directory in the cgroup filesystem (e.g., /sys/fs/cgroup/mygroup/).

2️⃣ Adding PIDs of processes you want to include into that cgroup by writing them to the file cgroup.procs:

echo <pid> > /sys/fs/cgroup/mygroup/cgroup.procs

3️⃣ The kernel then enforces any limits set in that cgroup on all the processes inside it, treating them as a single logical unit for resource control.

🔎 Practical grouping example

Imagine you have:

• A web server master process (PID 1000)
• Two worker processes (PIDs 1001, 1002)

To group them into a cgroup called myapp:

mkdir /sys/fs/cgroup/myapp/echo 1000 > /sys/fs/cgroup/myapp/cgroup.procsecho 1001 > /sys/fs/cgroup/myapp/cgroup.procsecho 1002 > /sys/fs/cgroup/myapp/cgroup.procs

Now:

✅ The kernel applies resource limits on myapp → collectively affecting all three processes.

📦 How does Kubernetes use cgroups?

This is where it all ties together beautifully:

🛠️ What Kubernetes does:

• Each container (e.g., a Docker container) gets its own cgroup on the node.