3 ms·
Because it does the rest for you, magically. You mount a config file into the system, but CoreOS on it (or boot per PXE), and coreOS does the entire rest for y
by kuschku 9y ago
Because it does the rest for you, magically.
You mount a config file into the system, but CoreOS on it (or boot per PXE), and coreOS does the entire rest for you. You never have to install or reinstall, never have to deal with updates. If you put that config file on any system with coreOS, it'll be exactly like yours.
This means I can create a config to tell a server to join a kubernetes cluster, and then whenever I create a new server I simply mount the config in, and am done. I never have to set up or configure servers. Everything is handled automatically. No more fighting with Ubuntu wiping all repos 10 minutes (!) after a version is replaced by the successor, never dealing with misconfiguration.
It's truly amazing, and I loved using it. Going back to normal distros is painful.
- quotemstr 9y agoAh, I see. So it's basically integrated Puppet, but with first class support from the configured system
- kuschku 9y agoYes, and this is fast. I can spin up a new server, have it configured, started, updated, and joined to my kubernetes cluster in 12 seconds (!). And it automatically handles updates across the cluster - only a certain amount of servers restarts at a time, allowing you to stay constantly up while handling updates automatically, and potentially scaling out in seconds. It's sad RedHat EOL'd it immediately.
- op00to 9y agoRed Hat EOLd what immediately?
- kuschku 9y agoOn the question if Red Hat will let container linux continue, the official answer was “Red Hat Enterprise Linux’s content, the foundation of our application ecosystem will remain our only Linux offering” TL;DR: container linux, as we know it today, is dead.
- alexandre_m 9y agoLike the above user said. No package management or bloated softwares distributed. It's immutable infrastructure. You configure your host once during provisioning, then it can accept container workloads in your cluster.