Recovery and auditing with innodb.fyi

Beyond inspection, inno helps when a tablespace is damaged or when you need to check a whole data directory at once.

Assessing what’s recoverable

• inno recover -f file.ibd - assess a damaged tablespace and count salvageable records.
• inno simulate -f file.ibd - predict how much would be recoverable at each innodb_force_recovery level (1-6), so you can choose a level deliberately instead of by trial and error.
• inno undelete -f file.ibd - attempt to recover delete-marked records, with a confidence score per record, and export them as CSV, JSON, or SQL.

Repairing safely

• inno repair -f file.ibd - recompute and write correct checksums. It creates a .bak backup by default; use --dry-run to preview and --no-backup to skip the backup.
• inno defrag writes a defragmented copy to a new file and never modifies the source.

Write operations keep a backup by default and can record an NDJSON audit trail, so recovery work stays traceable.

Auditing a directory

inno audit -d /var/lib/mysql/mydb scans an entire data directory for integrity problems, checksum mismatches, and index health, and can emit Prometheus metrics. To locate things across a directory, inno find searches for a page or scans for corruption, and inno tsid lists tablespace IDs.

A word of caution

Recovery tools operate on copies of damaged files whenever possible - always work from a backup of the original, never the only copy you have.

Facing a real recovery situation? Send us a message and a human will reply by email.