Docker: 'System has not been booted with systemd as init system (PID 1). Can't operate.'

System has not been booted with systemd as init system (PID 1). Can't operate. followed by Failed to connect to bus: Host is down is what systemctl prints inside a Docker container. A container does not boot an operating system, so there is no systemd to talk to. Do not try to start services; run the program itself in the foreground as the container's main process.

For a quick test inside a running container, service nginx start uses the old init script and works. For an image, make the server the CMD, for example CMD ["nginx", "-g", "daemon off;"], and let Docker start, stop and restart it.

The error

docker exec fix8-bld-sysd systemctl start nginx
System has not been booted with systemd as init system (PID 1). Can't operate.
Failed to connect to bus: Host is down

systemctl status nginx printed the same two lines, and both exited with code 1. The systemd package (255.4-1ubuntu8.18) was installed; the command exists, it just has nothing to connect to.

Why it happens

systemctl is only a client. It sends requests over D-Bus to systemd, which on a normal machine is process 1 and supervises every service. In a container, process 1 is whatever the image's CMD or ENTRYPOINT says. In the test container it was sleep 900; reading /proc/1/cmdline showed exactly that. No systemd, no bus, so systemctl refuses to operate.

This is by design: a container is meant to run one main process, and Docker plays the role of the service manager. Its restart policy, docker logs and health checks replace what systemd units do on a server. Running full systemd inside a container is possible with extra privileges and mounts, but it is rarely what you want.

The fix

FROM ubuntu:24.04
ARG DEBIAN_FRONTEND=noninteractive
RUN apt-get update && apt-get install -y --no-install-recommends nginx && rm -rf /var/lib/apt/lists/*
EXPOSE 80
CMD ["nginx", "-g", "daemon off;"]

docker run -d --name fix8-bld-web -p 127.0.0.1:15881:80 fix8-bld-p3fix
...
curl.exe -s -o NUL -w "%{http_code}`n" http://127.0.0.1:15881/
200

Process 1 in that container is nginx: master process nginx -g daemon off;, and the page answered 200. daemon off; matters: without it nginx forks into the background, process 1 exits and the container stops (the same image started with plain nginx showed Exited (0) three seconds later). For an interactive test in an existing container, service nginx start answered * Starting nginx nginx ...done. where systemctl had failed. Several services in one image call for one container each, wired together with Compose.

How it was reproduced

docker run -d --name fix8-bld-sysd ubuntu:24.04 sleep 900, then apt-get install of systemd and nginx inside it with docker exec, then systemctl start nginx. Ubuntu 24.04.5, systemd 255, nginx 1.24.0, Docker Desktop engine 29.9.2. The container, image and port are this test's own.

Frequently asked

Why does systemctl not work in a Docker container?

systemctl talks to systemd, which must be process 1. In a container process 1 is the image's CMD or ENTRYPOINT, so there is no systemd and no bus to connect to.

How do I start a service inside a Docker container?

Make the program the container's CMD and run it in the foreground, for example nginx -g daemon off. For a quick manual test, service name start uses the init script and works.

Should I run systemd inside Docker?

Usually not. Docker already restarts, logs and health-checks the main process. Run one service per container and connect several with Docker Compose.

More decoded errors in the Fixes category. To split an app and its database into containers, see Compose for an app and its database.