Ghost CLI migration rollback behavior after partial upgrade failure

Direct Answer

Ghost CLI does not perform an automated rollback when a database migration fails partway through an upgrade. Migrations that completed before the failure remain applied, and the CLI has no command to revert them. The operator must restore from the pre-upgrade SQL dump manually.

Confirmed Facts and Assumptions

These statements assume Ghost 5.x and are drawn from the supplied research brief, which is marked as requiring current verification. Treat them as planning context, not independently verified evidence.

• Ghost migrations run sequentially per file during an upgrade.
• A failure in a later migration does not automatically undo earlier migrations in the same batch.
• The migrations table records name, batch number, and timestamp only; it does not store checksums or rollback SQL.
• Ghost CLI creates a SQL dump before an upgrade, but restoration is manual, not orchestrated by the CLI.
• Custom migrations under core/server/data/migrations follow the same execution logic as core migrations.

Likely Explanation

The absence of rollback is consistent with a design where each migration file is an atomic unit, but the full upgrade run is not wrapped in a single reversible transaction. Because no rollback SQL or checksums are stored, the CLI cannot determine how to reverse a specific migration after it has been recorded as complete.

Recovery Steps for This Case

1. Stop the Ghost instance to prevent writes to the partially migrated database.
2. Inspect the migrations table to identify the last completed migration before the failure.
3. Locate the pre-upgrade SQL dump created by Ghost CLI immediately before the upgrade attempt.
4. Restore that dump using your database engine's native tool; the Ghost CLI does not orchestrate restoration.
5. Run ghost doctor to check environment consistency before retrying the upgrade.

Verification

mysql -u [user] -p [db_name] -e "SELECT * FROM migrations ORDER BY timestamp DESC LIMIT 10;"

Replace [user] and [db_name] with your database credentials. Confirm the dump file timestamp precedes the failed upgrade before restoring.

Missing Diagnostic Detail

Did the failure occur during a core Ghost migration or a custom migration under core/server/data/migrations? A custom migration may require manual cleanup of non-standard tables before a restore can succeed.