Integrating 3rd Party Containers in TripleO

Integrating 3rd Party Containers in TripleO

Building Containers

One of the following methods can be used to extend or build from scratch custom 3rd party containers.

Adding layers to existing containers

Any extra RPMs required by 3rd party drivers may need to be post-installed into our stock TripleO containers. In this case the 3rd party vendor may opt to add a layer to an existing container in order to deploy their software.

The example below demonstrates how to extend a container on the Undercloud host machine. It assumes you are running a local docker registry on the undercloud. We recommend that you create a Dockerfile to extend the existing container. Here is an example extending the cinder-volume container:

LABEL name="tripleo/centos-binary-cinder-volume-vendorx" vendor="Vendor X" version="2.1" release="1"

# switch to root and install a custom RPM, etc.
USER root
COPY vendor_x.rpm /tmp
RUN rpm -ivh /tmp/vendor_x.rpm

# switch the container back to the default user
USER cinder

Docker build the container above using docker build on the command line. This will output a container image <ID> (used below to tag it). Create a docker tag and push it into the local registry:

docker tag <ID>
docker push

Start an overcloud deployment as normal with the extra custom Heat environment above to obtain the new container.


Note that the new container will have the complete software stack built into it as is normal for containers. When other containers are updated and include security fixes in these lower layers, this container will NOT be updated as a result and will require rebuilding.

Building new containers with kolla-build

To create new containers, or modify existing ones, you can use kolla-build from the Kolla project to build and push the images yourself. The command to build a new containers is below. Note that this assumes you are on an undercloud host where the registry IP address is

Configure Kolla to build images for TripleO, in /etc/kolla/kolla-build.conf:


Use the following command to build all of the container images used in TripleO:

openstack overcloud container image build \
      --config-file /usr/share/tripleo-common/container-images/overcloud_containers.yaml \
      --kolla-config-file /etc/kolla/kolla-build.conf

Or use kolla-build to build the images yourself, which provides more flexibility and allows you to rebuild selectively just the images matching a given name, for example to build only the heat images with the TripleO customization:

kolla-build heat

Notice that TripleO already uses the /usr/share/tripleo-common/container-images/tripleo_kolla_template_overrides.j2 to add or change specific aspects of the containers using the kolla template override mechanism. This file can be copied and modified to create custom containers. The original copy of this file can be found in the tripleo-common repository.

The following template is an example of the template used for building the base images that are consumed by TripleO. In this case we are adding the puppet RPM to the base image:

{% extends parent_template %}
{% set base_centos_binary_packages_append = ['puppet'] %}

Integrating 3rd party containers with tripleo-heat-templates

The TripleO Heat Templates repo is where most of the logic resides in the form of heat templates. These templates define each service, the containers’ configuration and the initialization or post-execution operations.

The docker templates can be found under the docker sub directory in the tripleo-heat-templates root. The services files are under the docker/service directory.

For more information on how to integrate containers into the TripleO Heat templates, see the Containerized TripleO architecture document.

If all you need to do is change out a container for a specific service, you can create a custom heat environment file that contains your override. To swap out the cinder container from our previous example we would add:

    DockerCinderVolumeImage: centos-binary-cinder-volume-vendorx:rev1

3rd party kernel modules

Some applications (like Neutron or Cinder plugins) require specific kernel modules to be installed and loaded on the system.

We recommend two different methods to deploy and load these modules.

kernel module is deployed on the host

The kernel module is deployed on the base Operating System via RPM or DKMS. It is suggested to deploy the module via virt-customize. The libguestfs-tools package contains the virt-customize tool. Install the libguestfs-tools:

sudo yum install libguestfs-tools

Then you need to create a repository file where the module will be downloaded from, and uplaod the repo into the image:

virt-customize --selinux-relabel -a overcloud-full.qcow2 --upload my-repo.repo:/etc/yum.repos.d/

Once the repository is deployed, you can now install the rpm that contains the kernel module:

virt-customize --selinux-relabel -a overcloud-full.qcow2 --install my-rpm

Now that the rpm is deployed with the kernel module, we need to configure TripleO to load it. To configure an extra kernel module named “dpdk_module” for a specific role, we would add:

    dpdk_module: {}

Since our containers don’t get their own kernels, we load modules on the host. Therefore, ExtraKernelModules parameter is used to configure which modules we want to configure. This parameter will be applied to the Puppet manifest (in the kernel.yaml service). The container needs the modules mounted from the host, so make sure the plugin template has the following configuration (at minimum):

  - /lib/modules:/lib/modules:ro

However, this method might be problematic if RPMs dependencies are too complex to deploy the kernel module on the host.

kernel module is containerized

Kernel modules can be loaded from the container. The module can be deployed in the same container as the application that will use it, or in a separated container.

Either way, if you need to run a privileged container, make sure to set this parameter:

privileged: true

If privilege mode isn’t required, it is suggested to set it to false for security reaons.

Kernel modules will need to be loaded when the container will be started by Docker. To do so, it is suggested to configure the composable service which deploys the module in the container this way:

  command: /
    mode: "0755"
    content: |
      set -xe
      modprobe dpdk_module
          - {get_attr: [ContainersCommon, volumes]}
            - /var/lib/docker-config-scripts/

That way, the container will be configured to load the module at start, so the operator can restart containers without caring about loading the module manually.

Creative Commons Attribution 3.0 License

Except where otherwise noted, this document is licensed under Creative Commons Attribution 3.0 License. See all OpenStack Legal Documents.