Refactor roles to make use of quadlet instead of directly writing systemd services
Quadlet is the way going forward to run podman containers via systemd.
I will probably start by porting single container roles to it and after that my roles which are deployed to my monitoring host (caddy, monitoring etc.).
For single container roles this refactor is pretty straight forward. For multi-container roles (such as monitoring) this might be a bit more complex. I would have to either drop the pod approach and declare container networks or i would need to write kube.yml's and deploy those.
I'll probably go the former way for now as that would make it easier to restart single containers in case of an update.
Additionally quadlet has easy integration for the podman-auto-update service, which might make me go back to only pinning images by tag and not digest. Auto-update would then update the images on schedule if a new image is available. This would also cut down on the noise in this repo's PRs.
Roles TODO:
-
caddy -
grafana_agent -
monitoring -
uptime_kuma -
invoiceninja -
gitlab -
teamspeak -
ntfy -
keycloak -
vikunja -
wikijs -
camo -
cinny -
conduit -
scribbl -
libreddit -
gatus