A customer is being moved off a noisy-neighbour box to a quieter one because their bandwidth has outgrown the plan they’re on. You don’t need the Transfer Tool’s session scaffolding for one account; the pair of commands pkgacct and restorepkg does the job in a fraction of the time, and the failure modes are easier to reason about. The trade-off is that you lose the safety nets the Transfer Tool builds in by default, so the runbook has to add them back manually.
What’s happening
pkgacct on the source server builds a complete archive of the account — home directory, MySQL databases dumped to SQL, mail contents, DNS zone, user file, package reference — into /home/cpmove-<user>.tar.gz. restorepkg on the destination unpacks that archive and creates the account. Both commands are root-only; both produce archives that can be tested before commitment. What this path doesn’t give you: automatic DNS-cluster updates (you trigger them after the restore), automatic reseller-ownership reassignment (you set --owner= on restore), and automatic IP allocation (the restore uses the destination’s shared IP unless you specify).
How to verify
On the source, audit the account before packing:
whmapi1 accountsummary user=acme
du -sh /home/acme
whmapi1 list_dbs user=acme
pkgacct user=acme --skipbwdata --use-backup=no
ls -lh /home/cpmove-acme.tar.gz
pkgacct writes its progress to /usr/local/cpanel/logs/pkgacct.log. The archive lands in /home/; expect roughly 1.5x the home directory size because of the database dump. On the destination, before restore, confirm the username and domain are free:
whmapi1 list_users | grep acme
whmapi1 verify_user_domain_ownership domain=acme.com
The fix
- Transfer the archive between servers with rsync over SSH using key auth — never password auth, and never plain scp on a shared IP path:
rsync -avP /home/cpmove-acme.tar.gz root@dest:/home/. - Restore with explicit flags for owner, IP, and quota:
restorepkg --user=acme --owner=acmevpn --ip=203.0.113.42 /home/cpmove-acme.tar.gz. Omitting--ownerputs the account under root; omitting--ipuses the destination’s shared IP. - Update DNS before the customer notices the old IP returning NXDOMAIN. If the source and destination are in the same DNS cluster, propagation happens through cluster sync; if not, push the zone via
whmapi1 synczone domain=acme.comafter the restore. - Run a comparison check post-restore: source
accountsummaryversus destinationaccountsummary. The plan, quota, and email should match. The IP will differ; that’s expected.
For ongoing per-account moves between WHM servers, Stack Harbor’s cPanel server management team wraps pkgacct/restorepkg with the DNS propagation, monitoring updates, and customer comms steps so the move stays atomic from the customer’s perspective.