DuckDB Backup

DuckDB Backup Skill

Quick Start

Two backup methods, choose based on your needs:

File-Based Backup (cp method) - Fast for local backups

python3 scripts/​backup_duckdb.py \
  --db data/​awsntp.duckdb \
  --backup data/​backup-20251223.duckdb \
  --method cp

Native DuckDB Backup (attach method) - For remote/​cloud backups

python3 scripts/​backup_duckdb.py \
  --db data/​awsntp.duckdb \
  --backup data/​backup-20251223.duckdb \
  --method attach

With Automatic Timestamp Naming

python3 scripts/​backup_duckdb.py \
  --db data/​awsntp.duckdb \
  --backup data/​backups/ \
  --method cp \
  --timestamp
# Creates: data/​backups/​backup-20251223-153045.duckdb

Features

• Two backup methods: cp (fast, local) and attach (flexible, cloud-ready)
• Timestamped backups: Automatic unique naming with YYYYMMDD-HHMMSS format
• Error handling: Comprehensive logging and error reporting
• Scheduled backups: cron/​Task Scheduler integration for daily backups
• Verification: Tools to verify backup integrity and table counts
• Retention policies: Scripts to manage backup history and cleanup

When to Use

File-Based (cp method):

• Daily local backups (fastest approach)
• Pre-materialization safety snapshots
• Development/​testing environments
• When speed is critical
• Local machine backups

Native DuckDB (attach method):

• Cloud/​remote storage backups
• Cross-environment transfers
• Database in active use scenarios
• When you need portable backups
• Complex backup scenarios

Command Reference

python3 scripts/​backup_duckdb.py --db <source> --backup <target> [options]

Arguments:

• --db (required): Path to source DuckDB database
• --backup (required): Backup target (file path for cp, directory for attach with --timestamp)
• --method (optional): Backup method - cp (default) or attach
• --timestamp (optional): Add YYYYMMDD-HHMMSS timestamp to filename

Exit codes:

• 0: Backup successful
• 1: Backup failed (check logs)

Examples

Example 1: One-time Local Backup

python3 scripts/​backup_duckdb.py \
  --db ~/​LocalRepos/​awsntpdagster/​data/​awsntp.duckdb \
  --backup ~/​LocalRepos/​awsntpdagster/​data/​backup-20251223.duckdb \
  --method cp

Example 2: Daily Timestamped Backups to Folder

python3 scripts/​backup_duckdb.py \
  --db ~/​LocalRepos/​awsntpdagster/​data/​awsntp.duckdb \
  --backup ~/​LocalRepos/​awsntpdagster/​data/​backups \
  --method cp \
  --timestamp

Example 3: Scheduled Daily Backup (cron)

Add to crontab:

0 2 * * * python3 /​path/​to/​backup_duckdb.py --db /​path/​to/​awsntp.duckdb --backup /​path/​to/​backups/ --method cp --timestamp >> /​var/​log/​duckdb_backup.log 2>&1

Example 4: Backup with Retention Cleanup

#!/​bin/​bash
# Backup and keep only 7 most recent
BACKUP_DIR="/​path/​to/​backups"

python3 scripts/​backup_duckdb.py \
  --db data/​awsntp.duckdb \
  --backup "$BACKUP_DIR" \
  --method cp \
  --timestamp

# Keep only 7 most recent backups
ls -t "$BACKUP_DIR"/​backup-*.duckdb | tail -n +8 | xargs rm -f

Backup Strategy Selection

For 975 MB DuckDB database (awsntp.duckdb):

Advanced Usage

Verify Backup Integrity

# Check backup exists and is readable
duckdb data/​backup-20251223-153045.duckdb \
  "SELECT COUNT(*) as table_count FROM information_schema.tables;"

Compare Original and Backup Sizes

ls -lh data/​awsntp.duckdb data/​backup-*.duckdb | awk '{print $5, $9}'

List All Backups Chronologically

ls -lt data/​backups/​backup-*.duckdb | head -10

Troubleshooting

Issue: "duckdb: command not found"

• Install DuckDB CLI or update PATH
• Use full path to duckdb binary if attach method needed

Issue: "database is locked"

• Ensure no active connections to source database
• attach method can work around this (doesn't lock database)

Issue: "out of disk space"

• Check available disk space: df -h
• Use attach method for cloud storage
• Implement retention cleanup scripts

Issue: Slow backup

• For large databases, cp method is fastest
• Schedule backups during off-peak hours
• Consider incremental backup strategies

Integration with Your Pipeline

Pre-materialization Backup

# Backup before running expensive assets
python3 scripts/​backup_duckdb.py \
  --db data/​awsntp.duckdb \
  --backup data/​pre-materialization-backup.duckdb \
  --method cp

# Then run your asset materialization
dagster asset materialize -m awsntpdagster.definitions --select awsntp_features_merged_v6

Scheduled Daily Backup

For your awsntpdagster project:

macOS/​Linux crontab:

0 2 * * * cd /​Users/​zhaoliang/​LocalRepos/​awsntpdagster && python3 src/​awsntpdagster/​scripts/​backup_duckdb.py --db data/​awsntp.duckdb --backup data/​backups/ --method cp --timestamp

Reference

For detailed backup strategy guide, scheduling patterns, and retention policies, see:

• backup_strategies.md - When to use each method, cron setup, verification, troubleshooting

Performance Notes

• cp method: ~1-2 seconds for 975 MB database (your awsntp.duckdb)
• attach method: ~30-60 seconds for 975 MB database
• Both methods: Minimal CPU usage
• Disk space required: 1x original database size