uft-flocker-install cluster.yml fails/succedds randomly

[Stephenitis]

strange behavior I changed nothing and just re ran the command and it worked.

➜  coreos-test  uft-flocker-install cluster.yml
Attempting to connect to 52.91.100.67:22...
Connected to 52.91.100.67:22 after 0.00 seconds!
Attempting to connect to 52.91.169.59:22...
Connected to 52.91.169.59:22 after 0.00 seconds!
Attempting to connect to 52.91.195.228:22...
Connected to 52.91.195.228:22 after 0.00 seconds!
Attempting to connect to ec2-52-91-100-67.compute-1.amazonaws.com:22...
Connected to ec2-52-91-100-67.compute-1.amazonaws.com:22 after 0.00 seconds!
Enabled root login for 52.91.169.59
Enabled root login for 52.91.195.228
Enabled root login for 52.91.100.67
Retrying running ['-o', 'LogLevel=error', '-o', 'UserKnownHostsFile=/dev/null', '-o', 'StrictHostKeyChecking=no', '-i', '/host/Users/ClusterHQ/.ssh/oct-clusterhq-stephenitis.pem', '[email protected]', "bash -c 'echo; sudo mkdir -p /root/.ssh && sudo cp /home/core/.ssh/authorized_keys /root/.ssh/authorized_keys'"] on ec2-52-91-100-67.compute-1.amazonaws.com given result "Process exited with error code 1: \ncp: cannot create regular file '/root/.ssh/authorized_keys': File exists\n"...
Enabled root login for ec2-52-91-100-67.compute-1.amazonaws.com
Installed clusterhq-flocker-node on all nodes

#SECOND ATTEMPT
➜  coreos-test  uft-flocker-install cluster.yml
Attempting to connect to 52.91.100.67:22...
Connected to 52.91.100.67:22 after 0.00 seconds!
Attempting to connect to 52.91.169.59:22...
Connected to 52.91.169.59:22 after 0.00 seconds!
Attempting to connect to 52.91.195.228:22...
Connected to 52.91.195.228:22 after 0.00 seconds!
Attempting to connect to ec2-52-91-100-67.compute-1.amazonaws.com:22...
Connected to ec2-52-91-100-67.compute-1.amazonaws.com:22 after 0.00 seconds!
Enabled root login for 52.91.100.67
Enabled root login for 52.91.195.228
Enabled root login for 52.91.169.59
Enabled root login for ec2-52-91-100-67.compute-1.amazonaws.com
Installed clusterhq-flocker-node on all nodes

[lukemarsden]

I don't think the first attempt actually failed, it just logged its retry attempt and then succeeded. Should we make that more explicit in the output?