How do I rerun a failed migration?

Who can do this

Platform Admin, Infra Ops, or QA Admin (within scope)

Rerunning resumes only the tenants that failed in the previous attempt — successful tenants are left alone. The rerun is linked to the original job so the lineage stays intact.

Steps

1. Open Migrations and find the failed job (filter by Status: Failed).
2. Click the row to open the drawer.
3. In the drawer header, click Rerun.
4. The rerun dialog shows which tenants will be retried (the failed ones). Confirm.

Result

A new job appears with the same root job ID and an incremented attempt number. The job drawer shows the lineage — original attempt + every rerun — so you can compare logs across attempts.

For instance-replace

The rerun is resume-failed-only — it picks up exactly where the previous attempt stopped using the instance_replace_results artifact. Tenants that succeeded the first time are not re-imported.

Related

• How do I roll back a migration?
• How do I start a migration?