SDN pods CrashLoopBackOff with "Could not find an allocated subnet for node"

Issue

• During a scale up or new cluster install, the SDN pods on a few nodes are crashing with the following error:

I0512 19:55:06.890709   16017 cmd.go:160] Starting node networking (v0.0.0-alpha.0-59-g6e937ad0)
I0512 19:55:06.890730   16017 node.go:378] Starting openshift-sdn network plugin
W0512 19:55:06.895709   16017 subnets.go:159] Could not find an allocated subnet for node: ip-10-0-0-1.ec2.internal, Waiting...
W0512 19:55:07.901192   16017 subnets.go:159] Could not find an allocated subnet for node: ip-10-0-0-1.ec2.internal, Waiting...
W0512 19:55:09.406959   16017 subnets.go:159] Could not find an allocated subnet for node: ip-10-0-0-1.ec2.internal, Waiting...
W0512 19:55:11.662406   16017 subnets.go:159] Could not find an allocated subnet for node: ip-10-0-0-1.ec2.internal, Waiting...
W0512 19:55:15.044086   16017 subnets.go:159] Could not find an allocated subnet for node: ip-10-0-0-1.ec2.internal, Waiting...
W0512 19:55:20.112497   16017 subnets.go:159] Could not find an allocated subnet for node: ip-10-0-0-1.ec2.internal, Waiting...
W0512 19:55:27.711840   16017 subnets.go:159] Could not find an allocated subnet for node: ip-10-0-0-1.ec2.internal, Waiting...

• This is preventing the new nodes from starting, because the networking on the nodes isn't available yet.