This will hopefully avoid confusion (example) where people try to use their cluster from inside a Container and get errors because the cluster environment is not accessible in the container.
Now running anything but single-machine from inside the container will print out a (hopefully) very clear error message right away.
The very tiny minority of users who are able to hack things so that they somehow can use their cluster from inside a container will have to just add an CACTUS_INSIDE_DOCKER=0 to their command to override this behaviour.
This will hopefully avoid confusion (example) where people try to use their cluster from inside a Container and get errors because the cluster environment is not accessible in the container.
Now running anything but single-machine from inside the container will print out a (hopefully) very clear error message right away.
The very tiny minority of users who are able to hack things so that they somehow can use their cluster from inside a container will have to just add an
CACTUS_INSIDE_DOCKER=0to their command to override this behaviour.