Hi all!
As an evaluation of the complete Mender platform, I am trying some workflows as they would have been implemented in a production environment, which is not (yet) the case at this moment.
Using multiple of the docs, tutorials and blog posts, I created an easy device onboarding script which installs the mender-client4, mender-update etc.
The most important way of updating our software would be using a docker-compose file in which containers are added/removed or where container versions are altered.
Using the ‘Tutorial: Docker Compose update’, I was able to do a first test doing so using Mender. The tutorial walked me to all the steps smoothly, thanks for that!
However, I was wondering why the exact images included in the docker-compose file, also have to be mentioned in the app-gen command using --image parameter? I took this example from the tutorial:
app-gen
-- <other params>
-- image docker.io/library/traefik:v2.9 \ # This is mentioned already in the docker-compose file
-- image docker.io/traefik/whoami:latest \ # This is mentioned already in the docker-compose file
-- <other params>
To take it a step further, I also took a look at CI/CD integration. I have a working draft version which simply builds the docker-compose file and uploads the artifact to the Mender platform.
However, I had to hard-code the ‘–image’ parameters in the CI/CD yaml. For me to be useful, it would make sense if the --image parameters are dropped and the artifact is generated from the descriptions in the docker-compose alone.
Am I missing something obvious here?
Cheers!
