Unable to run systemd inside docker which is being run inside jenkins - docker

Question

I'm trying to get Jenkins to run Docker that runs SystemD.
So far I've been able to run systemd inside docker locally without Jenkins. Here are the steps to run it locally without jenkins:
# pull unop/fedora-systemd and create and run the container for it
sudo docker run --cap-add=SYS_ADMIN -e container=docker --tmpfs /run --tmpfs /tmp -v /sys/fs/cgroup:/sys/fs/cgroup:ro -t -i unop/fedora-systemd
# on a different terminal window, I can:
# get the container id of the "unop/fedora-systemd" image
sudo docker ps
# then exec bash on it
sudo docker container exec -t -i a98aa2bcd19e bash # where a98aa2bcd19e is the container id found above
# once inside the container, I can run systemd without any problems. examples:
systemctl status
systemctl start dbus.service
systemctl status dbus.service
The above works locally and I am able to run systemd inside the docker container.
The problem I get is when I try the same thing, but inside Jenkins.
I've tried to tweak Jenkinsfile several times, but not of my previous tries seemed to work. I always get an error when running under Jenkins similar to this:
+ systemctl status
System has not been booted with systemd as init system (PID 1). Can't operate.
Failed to connect to bus: Host is down
This is my latest Jenkinsfile that I've tried
pipeline {
agent {
docker {
image 'unop/fedora-systemd'
args '--cap-add=SYS_ADMIN -e container=docker --tmpfs /run --tmpfs /tmp -v /sys/fs/cgroup:/sys/fs/cgroup:ro -t -i'
}
}
stages {
stage('test') {
steps {
sh "echo hello world"
sh "systemctl status"
sh "systemctl start dbus.service"
sh "systemctl dbus.service"
}
}
}
}
On previous iterations of the Jenkinsfile, I've tried to replace -cap-add=SYS_ADMIN -e container=docker for --privileged, but that didn't help, I still got the same errors
Anyone have an idea of how can I get this to work? Why does the above work locally, but not on Jenkins? what am I missing here?
Note: Jenkins version: 2.150.2 and this is the Dockerfile used by unop/fedora-systemd
FROM fedora:rawhide
MAINTAINER http://fedoraproject.org/wiki/Cloud
ENV container docker
RUN dnf -y update && dnf clean all
RUN dnf -y install systemd && dnf clean all && \
(cd /lib/systemd/system/sysinit.target.wants/; for i in *; do [ $i == systemd-tmpfiles-setup.service ] || rm -f $i; done); \
rm -f /lib/systemd/system/multi-user.target.wants/*;\
rm -f /etc/systemd/system/*.wants/*;\
rm -f /lib/systemd/system/local-fs.target.wants/*; \
rm -f /lib/systemd/system/sockets.target.wants/*udev*; \
rm -f /lib/systemd/system/sockets.target.wants/*initctl*; \
rm -f /lib/systemd/system/basic.target.wants/*;\
rm -f /lib/systemd/system/anaconda.target.wants/*;
VOLUME [ "/sys/fs/cgroup", "/tmp", "/run" ]
CMD ["/usr/sbin/init"]
PS: I've seen a related question, but what they were asking is different

Answer 1

I did not know about the related question. Let me point out again that you do not need to run a systemd daemon in a systemd controlled container if it is just about running multiple services in it. Simply overwrite /usr/bin/systemctl with the docker-systemctl-replacement script. Then go to register it with CMD ["/usr/bin/systemctl"] as the init process of the container.
That's it. Now you can run any systemctl-start process from the operating system. It works to the extent that even provisioning with ansible/puppet scripts have no problem at all. An specficially, I am using that to provision Jenkins images with the operating system that the developers like to have as a basis. No priviledged mode required.

Answer 2

You may try an image that has Fedora with System D already active with this command:
docker run -d --name systemd-fedora --privileged -v /sys/fs/cgroup:/sys/fs/cgroup:ro jrei/systemd-fedora
Then you just need to run:
docker exec -it systemd-fedora /bin/bash
and there you can just install, start and restart any service you need.