Daily export of Hasura metadata to a JSON file on disk.

nself backup hasura-metadata calls the Hasura export_metadata API and writes the result to:

<backup_dir>/hasura-metadata-<YYYY-MM-DD>.json

The file is mode 0600. The backup directory defaults to ./backups (relative to the project directory) or the value of NSELF_BACKUP_DIR.

Hasura metadata contains all tracked tables, relationships, permissions, remote schemas, and event triggers. Losing it means manually re-applying every schema change. A daily JSON snapshot makes recovery a single nself db hasura metadata apply call.

nself backup hasura-metadata

Example output:

hasura metadata backup written to ./backups/hasura-metadata-2026-04-30.json

Install the system-level cron (fires at 02:00 UTC):

# Linux (systemd) or macOS (launchd) — requires root
sudo nself maintenance hasura-metadata-cron --install

Check the status:

nself maintenance hasura-metadata-cron --status

Remove the cron:

sudo nself maintenance hasura-metadata-cron --remove

On Linux the timer is named nself-hasura-metadata-backup.timer (systemd). On macOS the LaunchDaemon label is org.nself.hasura-metadata-backup.

Logs:

# Linux
journalctl -u nself-hasura-metadata-backup

# macOS
tail -f /var/log/nself-hasura-metadata-backup.log

nself doctor --deep includes the check Hasura metadata backup (BACKUP-METADATA-01).

Fix a fail result:

nself backup hasura-metadata

# Copy the backup JSON into the project's hasura/ directory.
cp backups/hasura-metadata-2026-04-30.json hasura/metadata.json

# Apply it to a running Hasura instance.
nself db hasura metadata apply

Set NSELF_BACKUP_DIR in .env.local or .env.secrets to override the default ./backups path:

NSELF_BACKUP_DIR=/var/backups/nself

• cmd-backup — full backup command reference
• operations/release-cascade — release checklist that includes a metadata backup step
• Home