innodump
An offline reader for InnoDB tablespace files (.ibd, ibdata1) across MySQL 5.6–8.4 and MariaDB — reconstructs CREATE TABLE statements and extracts row data via direct B+tree traversal, without a running database server.
innodump reads InnoDB tablespace files — file-per-table .ibd files or the shared ibdata1 — directly off disk, reconstructing CREATE TABLE statements from embedded SDI dictionary information or .frm files, then walking the B+tree to extract row data. It handles every InnoDB row format (DYNAMIC, COMPACT, REDUNDANT, COMPRESSED) and understands INSTANT ADD/DROP COLUMN history, and can recover deleted-but-not-yet-purged rows with a --deleted-only flag. Output is either SQL INSERT statements or MySQL Shell-compatible TSV, for a single file or recursively across a directory, with a --skip-corrupted option to tolerate damaged pages.
It’s explicitly read-only and, per the author, “not a backup tool” — it’s meant for consistent snapshots (a stopped server, FLUSH TABLES ... FOR EXPORT, or an existing backup), offering best-effort data recovery from files that are no longer attached to a live server, not a substitute for one that is. Written in Go with no external dependencies, GPL-3.0-or-later licensed.
Community feedback
This space is for grounded, evaluable experience — versions used, environments, what you measured or observed, and how it behaves alongside other software. General impressions without supporting detail may be removed. Sign in to write one; posts are tied to a verified identity, not anonymous.