## TL;DR
The sqlite session DB got half-written, usually from a killed run or a full disk. Confirm the corruption with sqlite3, delete the file, restart with `--flush-session`, and back up session files before killing long runs in the future.

```text
unable to resume previous session
```

## Use this when
- Sqlmap refuses to resume, citing a corrupt or unreadable session file.
- The previous run was killed, crashed, or the disk filled.
- You want to continue the same engagement's testing.

## Not for this skill when
- You actually want a fresh session. Just use `--flush-session` and move on.
- The session file belongs to a different target. Start a new session for the new target.
- Sqlmap is a different version than the one that wrote the file. Version skew can fake corruption.

## Steps
1. Confirm the corruption: `sqlite3 ~/.sqlmap/output/target/session.sqlite ".tables"`. Verify: sqlite errors instead of listing tables.
2. Check disk space: `df -h`. A full disk is the usual corruptor. Verify: space is available now.
3. Remove the corrupt file: `rm ~/.sqlmap/output/target/session.sqlite`. Verify: the file is gone.
4. Rerun with `--flush-session` to start clean. Verify: sqlmap begins a new session without the resume error.
5. For long runs, copy the session file to a backup before killing the process. Verify: the backup opens in sqlite3.

## Variant phrasings
### sqlmap session corrupt resume
Compact phrasing.
### sqlmap unable to resume session
Alternate wording.
### sqlmap session sqlite corrupt
The storage-level search.

Compatibility: sqlmap 1.x stores sessions as sqlite under `~/.sqlmap/output/`. Paths differ on Windows.

## Why it happens
The session DB is written continuously during the run. A SIGKILL, power loss, or full disk interrupts a write mid-transaction, leaving the file in a state sqlite will not open. Sqlmap then cannot resume and tells you so.

## Edge cases / pitfalls
- Never delete the whole output directory in frustration; the log files in it are engagement evidence. Delete only the session sqlite file.
- Resuming across sqlmap versions can corrupt or misread sessions. Keep one sqlmap version per engagement.
- If corruption recurs without kills, suspect the disk or filesystem. Run a disk check before blaming sqlmap.
- A backup taken mid-run is only as good as its last write. Note the timestamp so you know what it covers.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_HA4qAyCauU6ebDYwWR-pow