Post #4354488
2026-08-03 11:57 UTC
Replies (1)
-
@oliof@social.treehouse.systems 2026-08-03 12:26
@sushee@ohai.social I remember back in the day when @isotopp@infosec.exchange, me, and a couple other people decided to get off of upstream puppet modules because those had way too many eventualities that never applied. To save us from accidentally having some slip in we chose Ansible instead as well as a short number of rules how to construct our scripts ('no configuration generated via yaml indirections', 'service deployment and runtime management are different tasks that need to be addressed independently' and two or three more), and we managed to rebuild from scratch in a couple weeks what previously was 9 months of fighting upstream changes and dependency hells for aspects of our services that never applied to our use case. Allowing yourself to simplify instead of being dragged down by the baggage of 95% of never-used complexity is super important.