I’m upgrading RancherOS, in place, on one of my nodes in the my DEV cluster and frankly it’s a fail. What started off as a quick download has turned into a build. As to what Rancher is building on my system I have no idea, however, when compared to CoreOS auto deploy interruptions I do not care. I can deal with a system that is rebooted and the Docker Swarm can too… but having to wait while Rancher recompiles swaths of the system and causes the VMware web interface to crash or timeout.
Furthermore with a 32GB ram system having to partition the VMware by node I’ve already discovered some applications unable to complete the task because the cap on memory allocation.