We need to consider whether it would be advantageous for a node to be able to boot up on an image where the network configuration cannot be found for the node that's booting up. Today, this fails and the node drops to the emergency shell:
Perhaps (and this really is meant to be a question) we should default back to failing more gracefully and attempting DHCP instead, that way new nodes that may not have been known about when the original image was created could be used in the wild, albeit requiring post-deployment network configuration, otherwise this forces image rebuilding for all new nodes, which may not always be desired.
(We'd also need to consider how this would impact the placement of # combustion: network in the main combustion script.)
We need to consider whether it would be advantageous for a node to be able to boot up on an image where the network configuration cannot be found for the node that's booting up. Today, this fails and the node drops to the emergency shell:
Perhaps (and this really is meant to be a question) we should default back to failing more gracefully and attempting DHCP instead, that way new nodes that may not have been known about when the original image was created could be used in the wild, albeit requiring post-deployment network configuration, otherwise this forces image rebuilding for all new nodes, which may not always be desired.
(We'd also need to consider how this would impact the placement of
# combustion: network
in the main combustion script.)