make TOAD more docker environment for real-use case.

[s1061123]

For now, TOAD's docker is just for testing and not matured as docker container environment.

We need to work on following items:

• identify user data (jjb files or some) and store it in locally (mount with '-v' option)
• make clear docker build pipeline
• make docker container image and put it to dockerhub
• stop systemd and open only minimum port with EXPOSE in dockerfile

[zshi-redhat]

identify user data (jjb files or some) and store it in locally
@s1061123 Is this the reason that JJB files are not loaded automatically with docker method after synchronizing JJB config ?

[s1061123]

@zshi-redhat it seems not to related. Could you please file it in new issue ticket?

[s1061123]

For user data, especially JJB files, we need to take care carefully because ansible-playbook can overwrite it in next provision.
we need to add some tags at copying JJB file in docker case, like 'docker_copy_jjb' then we could skip it with "--skip-tags docker_copy_jjb".

[s1061123]

#83 adds mapping container data to host directory.

[leifmadsen]

Honestly I think there is a possible need that we want to start breaking out configuration into locally deployed directories, and then mounting those into the image, rather than writing directly to the image. Of course we then need to determine if we're going to continue to support the OpenStack deployment method as well, or say that containers are the only way we're going to allow deployment of TOAD.

If we continue to support the OpenStack method, then I think we need to keep "doing it wrong" on the container side of things and configuring the image. If we do that, then I think the container image will really just be a data image that still needs to be "configured" via Ansible (same as we do now), but we'll be able to skip downloading anything from the container image, as all binaries will already be installed.

Thoughts?

[dougbtv]

Honestly I think there is a possible need that we want to start breaking out configuration into locally deployed directories, and then mounting those into the image, rather than writing directly to the image.

Just wanted to say that I like this approach.

Unsure about the OpenStack method, definitely defer to others use-case on that one.

[dougbtv]

Also thought I'd mention we might be able to use docker cp to copy files into a docker volume, fwiw.

More info @ https://docs.docker.com/engine/reference/commandline/cp/