The servers became unreachable, but the IPs weren't unreachable until all the servers were reconfigured.
My test should have caught this bug:
> In this event, the canary step correctly identified that the new configuration was unsafe. Crucially however, a second software bug in the management software did not propagate the canary step’s conclusion back to the push process, and thus the push system concluded that the new configuration was valid and began its progressive rollout.