Peek inside a container process without entering its shell
Container Basics On Linux (Namespaces/Cgroups)
Peek inside a container process without entering its shell
🧩 The Challenge
Trying to debug what a container is actually doing often leads to installing SSH or net-tools inside the image, which is a massive security headache. I’ve wasted hours cleaning up bloat just because I needed to see a process tree.
💡 The Fix
Use nsenter to borrow the namespace of the running container process from the host. It lets you run binaries from your host tools directly in the container’s isolated environment.
PID=$(docker inspect -f '{{.State.Pid}}' <container_name_or_id>)
nsenter -t $PID -n -p ps aux
⚙️ Why It Works
By grabbing the PID of the container’s init process, you’re telling the kernel to switch your own session into that process’s network and process namespaces. You effectively hijack their view of the system without them ever knowing you’re there.
🚀 Pro-Tip: Alias this as “ns-exec” and save yourself the headache of memorizing namespace flags.
Linux Tips & Tricks | © ngelinux.com | 11/17/2026
