# HiveMQ Cloud Serverless cutover checklist

Updated October 20, 2026. Complete migration before December 31, 2026.

- [ ] Record devices, backend consumers, connections, rate, payload sizes, fan-out and subscriptions.
- [ ] Identify hard-coded hostnames, certificate pins and offline devices.
- [ ] Check target plan capacity, retained/QoS needs, burst limits, region, support and SLA requirements.
- [ ] Recreate device credentials and least-privilege policies; verify allowed and denied operations.
- [ ] Verify hostname, TLS port or full WSS URL and certificate trust on representative clients.
- [ ] Run the local verifier; require actual receipt, not just a publish acknowledgement.
- [ ] For retained-dependent workloads, require correct replay and retain packet flags; investigate the documented RunMQTT limitation.
- [ ] Plan retained snapshots/current-state republishing, session recreation and durable recovery of queued work.
- [ ] Coordinate command publishers and consumers; handle duplicate business events.
- [ ] Test restart, network interruption, offline recovery, LWT and peak startup traffic.
- [ ] Move one batch, observe application delivery and actual usage, then decide whether to expand.
- [ ] Keep a timed rollback window while the source is still available.
- [ ] Accept every required device and backend before deleting the source. Deleted Serverless clusters cannot be recreated.
- [ ] Clear synthetic retained state, revoke test identities and delete local credential files.

Guide: https://runmqtt.com/docs/migrate-from-hivemq-cloud
Official notice: https://docs.hivemq.com/hivemq-cloud/serverless-end-of-life.html