I'm interested in moving to Kubernetes to make use of the templating languages available that are better than plain Ansible jinja2, and also offer features like schema checking.
Because my services are pretty integrated together and to avoid having hardcoded values in multiple places my Ansible files are a pain to maintain
The pain of Ansible+Jinja2 is what eventually pushed me to write judo[1]... It works surprisingly well for what it was built for, but of course has limitations (e.g. there's no direct support for a templating language; you just plug env variables into a script). The idea is, do less, and allow other tools to fill the gaps.
There's still a lot on my todo list, like env files, controlling parallelism, canaries/batches, etc. I'm currently doing these things using hacky shell scripts, which I don't like, so I'd prefer moving that into the main binary. But I still prefer it as-is over Ansible.
I'm interested in moving to Kubernetes to make use of the templating languages available that are better than plain Ansible jinja2, and also offer features like schema checking.
Because my services are pretty integrated together and to avoid having hardcoded values in multiple places my Ansible files are a pain to maintain