Skip to content

Decommissioning a cPanel server — the structured shutdown

How to safely take a cPanel server out of service: data migration, customer communication, license return, and the final wipe.

Decommissioning a cPanel server gets less attention than provisioning one, and it shows. The typical mistake is “we’ll just stop billing for it” — and a year later, the server is still running because nobody can confirm everything was moved off, the license is still being paid, and someone’s customer has a forgotten DNS pointing at it.

The trade-off

A rushed decommission risks data loss and customer outage. A thorough one is days of structured work. The middle path — explicit checklist, defined timeline, no shortcuts — is what works.

How to verify — the decommission checklist

T-30 days — announce and inventory:

  1. List all accounts on the server. whmapi1 listaccts
  2. List all domains served (including parked, addon, subdomains). Cross-reference with DNS records.
  3. List all mail accounts and current mailbox sizes.
  4. Identify any backups stored on this server that are destinations for other systems.
  5. Notify customers of the decommission date in writing.

T-21 days — migrate: 6. Migrate accounts using WHM Transfer Tool to the destination. See the migration patterns framing. 7. Migrate mail using imapsync to ensure UIDL continuity. 8. Update DNS to point at the destination, with TTL pre-lowered to ~300s.

T-7 days — verify and snapshot: 9. Confirm all expected accounts have migrated, with correct disk usage, mail folder counts, and database row counts. 10. Take a full backup of the source server. Verify it can be read on the destination. Store on append-only off-site for 90 days minimum. 11. Confirm DNS has propagated and traffic is hitting the new server. Watch logs on the old server — anything still hitting it is a misconfigured client.

T-1 day — final cutover: 12. Set the source server to read-only via firewall rules; allow only management traffic. 13. Verify no customer traffic for the last 24 hours. If there is, investigate before continuing.

T-0 — shutdown: 14. Stop services on the source. 15. Take a final block-level snapshot (in case of “we forgot something”). 16. Verify the source is offline and confirm the destination is handling all expected traffic.

T+30 days — finalize: 17. Return the cPanel license (per cPanel’s process; otherwise you keep paying for it). 18. Return CloudLinux, LiteSpeed, JetBackup, and any other per-server licenses. 19. Securely wipe disks if hardware is being returned or repurposed. NIST 800-88 if the hardware ever held customer data. 20. Archive the decommission report — which accounts moved where, which licenses were returned, which DNS records were updated.

What to do

  1. Follow the checklist. The “we’ll skip step X” path is how data gets lost.
  2. Don’t rush the announcement window. Customers need time to notice and react.
  3. Verify, don’t assume. Each migration step has a verification step; do them.
  4. Document everything. Future you investigating “what happened to that server” will thank present you.

Stack Harbor handles decommissioning under migration services when retiring a server is part of a broader move.