| assets | ||
| cmd | ||
| config | ||
| docker | ||
| docs | ||
| Godeps | ||
| init | ||
| scripts | ||
| util | ||
| .dockerignore | ||
| .drone.yml | ||
| .gitignore | ||
| .package | ||
| .wrap-docker-args | ||
| build.sh | ||
| main.go | ||
| README.md | ||
RancherOS
It's containers all the way down! Everything is powered by Docker (I mean everything).
Building
Docker 1.5+ required.
./build.sh
Running
KVM, VirtualBox, and VMware all supported (Clouds and Vagrant coming soon). Just
boot from the rancheros.iso (probably want to run with at least 1G of memory) from the releases page.
Logging in
Log in with rancher/rancher and use sudo to get root access.
Once cloud-init integration is finished we will not need to hard code the password anymore.
Persisting State
Create a partition with the label RANCHER_STATE, for example
mkfs.ext4 -L RANCHER_STATE /dev/sda
Configuring
The entire state of RancherOS is controlled by a single configuration document.
You can edit the configuration with the rancherctl config command.
Commands
docker -- Good old Docker, use that to run stuff.
system-docker -- The docker instance running the system containers. Must run as root
rancherctl -- Control and configure RancherOS
How does this work
Everything in RancherOS is a Docker container. We accomplish this by launching two instances of
Docker. One is what we call the system Docker which runs as PID 1. System Docker then launches
a container that runs the user Docker. The user Docker is then the instance that gets primarilry
used to create containers. We created this separation because it seemed logical and also
it would really be bad if somebody did docker rm -f $(docker ps -qa) and deleted the entire OS.
