## Simpler dependencies - No docker swarm? How will we restart unhealthy containers then? - No EFS? - No Terraform? ## Logic - Move descriptor logic from management-service here - Move service upgrade rolling upgrade from management-service here - No need for bootstrap management-service circular flow ## Recovery - Failsafe mechanism for doomsday recovery - Based on cron that downloads and executes a signed shell script / executable - This will be the future mechanism which we upgrade Boyar with - This should be the first feature deployed - through BoyarV2 upgrade mechanism ## Monitoring and long term stability - Monitor Dockerhub rate limits (so node IP will not be blacklisted for 6 hours) - Prune old images from cache to avoid running out of space - Prune old files from root device to avoid running out of space - Monitor machine disk space / memory / CPU (average not spot)
