a (hopefully not too poor) attempt at making a self-operating etcd cluster in docker swarm

https://github.com/YouMightNotNeedKubernetes/etcd Most code is taken from this repo, I'm just trying to bend etcd into not having to actively manage it. The organization has been recently archived and it's being migrated to https://github.com/swarmlibs .

• still under initial development, started the repo to have an easier time testing it on play-with-docker
• not fit for production unless you know what you're doing
• it's completely ignoring what happens when the etcd cluster grows in size of data, it's meant to be kept small for the specific use case

• single command startup
• no commands should be run manually against the etcd API to manage membership
• should be a 1-to-1 mirror of which nodes are labeled to run the etcd cluster
• resilient to node failure
• shouldn't care on which nodes it is actually running
• error cases when joining cluster should be handled automatically