Migration fails when teams switch everything at once. A staged cutover keeps risk low and service continuity high.
Migration plan
- Mirror current prompts and channel config in ClawStack.
- Validate with test traffic.
- Run dual environment for a short overlap window.
- Route new users to ClawStack first.
- Drain legacy VPS traffic and decommission.
Data and secret handling
- Rotate gateway and provider tokens during migration.
- Keep secrets out of migration notes and tickets.
- Audit env vars before cutover.
Rollback criteria
- Define explicit failure threshold (latency/error rate).
- Keep VPS runtime warm until success window closes.
- Document rollback command path in advance.
Post-cutover tasks
- Remove legacy DNS/ports.
- Archive old logs and snapshots.
- Update runbooks and onboarding docs.
A structured migration usually takes less time than prolonged VPS troubleshooting and gives cleaner operational controls afterward.