The cluster that would not start, because 1000000 stopped being a number

New Linode Kubernetes clusters stayed pending after subnet IDs crossed 1,000,000. This explains why an ID must not pass through a float, and shows Go and Python guards plus a boundary test.