The GitHub Actions UTC trap is the one I see bite teams most: a schedule that works in a local crontab silently shifts when copied into a workflow, and the same applies to Kubernetes CronJobs before spec.timeZone landed. One check worth adding to the YAML checker for CronJobs is startingDeadlineSeconds together with concurrencyPolicy — if the controller misses a run by more than that deadline the job is skipped, and with Allow concurrency a delayed run can pile up duplicate jobs instead. The VPC arithmetic is spot on too: planning on 251 usable addresses per /24, not 256, saves a painful re-plan later.