kinvolk/kube-spawn

Starting kube-spawn fails if the images are already there

Closed this issue · 0 comments

Trying to start kube-spawn when the (cloned) node images are present results on errors.

# ./kube-spawn start
using config from /var/lib/kube-spawn/default
using "coreos" base image from /var/lib/machines
spawning cluster "default" (3 machines)
Image resized.
Resize '/var/lib/machines' of 'max'
waiting for machine "kubespawndefault2" to start up
waiting for machine "kubespawndefault0" to start up
waiting for machine "kubespawndefault1" to start up
failed to start machine: error cloning image: "/bin/machinectl clone coreos kubespawndefault0" failed: Could not clone image: File exists
failed to start machine: error cloning image: "/bin/machinectl clone coreos kubespawndefault1" failed: Could not clone image: File exists
failed to start machine: error cloning image: "/bin/machinectl clone coreos kubespawndefault2" failed: Could not clone image: File exists
not all machines started

The current workaround is running

$ sudo machinectl remove kubespawndefault{0,1,2}

I think we should remove them ourselves, or reuse them.