This project is for generating a PXE server for different environments. It uses Jinja2 templates and context files to generate the specific configuration and kickstart files for the pxe server. The image is built with most of the popular pxe target images built in.

Besides docker, there are a few python dependencies. You can install them using pip and the requirements.txt file.

pip install -r requirements.txt

There’s an example context included. To generate the associated files, you can execute

make pao19-centos

You’ll probably see something like

rm -f confs/*
rm -f http_files/*
./bin/render dnsmasq.conf pao19-centos > confs/dnsmasq.conf
./bin/render dnsmasq.hosts pao19-centos > confs/dnsmasq.hosts
./bin/render nginx.default pao19-centos > confs/nginx.default
./bin/render resolv.dnsmasq pao19-centos > confs/resolv.dnsmasq
./bin/render default pao19-centos > confs/default
./bin/render centos.cfg pao19-centos > http_files/centos.cfg

To add a new environment, you have to edit the Makefile until I figure out how to fully generalize the targets. For example, if we were adding a new environment called ‘docker-cluster’, we’d

1. Create a new context in contexts/docker-cluster.yml

2. Create a new make target in environments/. If making one for an environment called ‘docker-cluster’, you might create a file called environments/docker-cluster.mk that looks like

   docker-cluster: CONTEXT=docker-cluster docker-cluster: clean files cfg_files

Currently, there are two types of templates. Ones for files served over http to clients and ones that are copied over to the container. Both of the kinds of templates are placed in the ‘templates’ directory. So step one is just creating the template itself and placing it there.

Step two is creating the target in the Makefile. For new http served templates, add them to the https: target. For others, add to the files target. Should be pretty obvious from looking at the current Makefile.

Once you’ve generated the configuration files with

make <environment_name>

The container can be built. For example,

docker -t jameskyle/pxe-server .

The server runs as the dnsmasq user and needs access to the hosts network. The following commmand provides those capabilities.

docker run -d --net=host --privileged=true --name pxe-server \
jameskyle/pxe-server /sbin/my_init

If you want to debug the server, you can run a bash shell

docker run --net=host --privileged=true -t --name pxe-server -i \
jameskyle/pxe-server /sbin/my_init -- /bin/bash