13-mongodbTermsLevel_10mongodump / mongorestore (Backups)

mongodump / mongorestore (Backups)

mongodump / mongorestore (Backups)

Level 10 — Administration, Security & Advanced Features The command-line database backup and recovery utilities that export collections into binary BSON and metadata JSON files (mongodump) or import them back to recreate collections and rebuild indexes (mongorestore), serving as the equivalent to PostgreSQL's pg_dump and pg_restore.

1. Prerequisites

2. Term Category

Administration / Operations (BSON Backup & Restore Utilities): mongodump and mongorestore are CLI utilities for creating compressed BSON binary database backups and restoring collection snapshots.

3. Explanation

Environment Context

  • Operating System Shell (Standalone terminal executable commands run in the bash/linux console. They are not run inside the mongosh database shell).

(1) Design Motivation — "Why did we design this?"

To protect against system crashes, administrator errors, or ransom attacks, you must maintain backups of your database.

In PostgreSQL, you export tables using: pg_dump -U username dbname > backup.sql

We designed mongodump and mongorestore to perform this task in MongoDB.

They are high-performance utility tools.

Instead of exporting data to slow, human-readable JSON files (which lose BSON data type information like specific Int vs. Double bounds), mongodump streams the raw binary BSON blocks from the database directly onto your disk files.

This guarantees that data types, binary files, and indexes are preserved exactly.

(2) The Export Formats

When you run mongodump, it creates a directory containing two file types per collection:

  1. .bson Files: The raw binary BSON document bytes.
  2. .metadata.json Files: The index definitions and collection settings.

When mongorestore reads this directory, it uses the .bson files to populate the documents and automatically reads the .metadata.json files to rebuild all collection indexes, saving you from having to recreate indexes manually.

(3) Reality Metaphor (Office Photocopying)

Imagine backing up a paper medical files archive room:

  • mongodump: You make photocopies of all client folders, bind the sheets into cardboard boxes (the BSON files), and write down the shelf names and index sorting rules in a notebook (the metadata JSON).
  • mongorestore: You take those cardboard boxes to an empty room.
    • You read the notebook to install matching shelves and index dividers (index rebuild).
    • You place the sheets back into the drawers.
    • The filing system is restored.

(4) Code Examples

Running Backups in the Linux Terminal

These commands are run in your standard shell, not inside mongosh:

# 1. Back up the 'shop' database, compressing the output with gzip
mongodump --db shop --out /backups/july-2026 --gzip

# This creates a folder: /backups/july-2026/shop/
# containing user.bson.gz, user.metadata.json.gz, etc.

# 2. Restore the backup to a local database, rebuilding all indexes
mongorestore --db shop_restore /backups/july-2026/shop --gzip

4. Common Mistakes & Pitfalls

Mistake 1: Attempting to execute the 'mongodump' or 'mongorestore' commands inside the 'mongosh' javascript shell prompt

The mistake: Connecting to the database via mongosh and typing mongodump --db shop inside the database prompt, resulting in syntax and reference errors.

Why it's wrong: mongodump and mongorestore are standalone, compiled binary executables managed by the operating system shell.

They are not JavaScript functions, so the mongosh interpreter does not understand them.

Fix: Exit the mongosh shell (or open a new terminal tab) and run mongodump directly in your operating system command line prompt.

Mistake 2: Using mongodump for Multi-Terabyte Live Production Database Backups

The mistake: Running mongodump against a 5TB production cluster during business hours.

Why it's wrong: mongodump executes query scans across all collections, causing heavy CPU and WiredTiger cache pressure. For large databases, use Atlas Continuous Backups or LVM/EBS Volume Snapshots.

Incorrect:

$ mongodump --uri "mongodb://prod:27017" # ❌ Scans 5TB live database!

Fix:

Use Cloud Storage Volume Snapshots (AWS EBS / GCP Persistent Disk) or Atlas Backups

Mistake 3: Restoring Dumps with mongorestore Without Specifying --drop Flag on Existing Collections

The mistake: Restoring a dump file into a database containing partial legacy data without --drop.

Why it's wrong: Without --drop, mongorestore merges dump records into existing collections, causing E11000 duplicate key collisions.

Incorrect:

$ mongorestore --db app dump/app # Merges into existing dirty collection

Fix:

$ mongorestore --db app --drop dump/app # Drops existing collection before restore

5. Practice Exercises

Exercise 1: Exporting Database Backups with mongodump

Scenario: Export a compressed BSON backup of database store_db to directory /backups/2026-08-05 using mongodump.

Requirements:

  1. Execute mongodump --db=store_db --out=/backups/2026-08-05 --gzip.
Answer

Implementation

mongodump --uri="mongodb+srv://user:pass@cluster0.abc.mongodb.net/store_db" >   --out=/backups/2026-08-05 >   --gzip

Technical Explanation

  1. mongodump exports collection data as raw binary BSON files alongside JSON metadata definitions.
  2. --gzip compresses output files to minimize backup disk storage.
  3. Standard CLI utility for logical database backups.

Exercise 2: Restoring Database Backups with mongorestore

Scenario: Restore compressed database backup /backups/2026-08-05/store_db into target database store_db_restored using mongorestore.

Requirements:

  1. Execute mongorestore --db=store_db_restored --gzip.
Answer

Implementation

mongorestore --uri="mongodb://localhost:27017" >   --db=store_db_restored >   --gzip >   /backups/2026-08-05/store_db

Technical Explanation

  1. mongorestore parses BSON backup files and inserts documents and secondary indexes into target collections.
  2. --nsInclude or --db allows restoring backups under new database namespace names.
  3. Core command for disaster recovery and staging environment populating.

Exercise 3: Point-in-Time Oplog Backup Restoration

Scenario: Perform a point-in-time backup restoration using mongodump --oplog and mongorestore --oplogReplay.

Requirements:

  1. Explain --oplog capture during logical dumps.
Answer

Implementation

# 1. Capture dump with active oplog
mongodump --uri="..." --oplog --out=/backups/pit_dump

# 2. Restore with oplog replay to capture writes that occurred during dump execution
mongorestore --uri="..." --oplogReplay /backups/pit_dump

Technical Explanation

  1. --oplog captures oplog entries generated while mongodump was running.
  2. --oplogReplay replays those oplog entries after restoring BSON files, ensuring point-in-time transactional consistency.
  3. Prevents dirty backup state in high-write production databases.

7. Key Takeaways

  • mongodump exports databases to disk; mongorestore restores them.
  • Direct NoSQL equivalent to SQL's pg_dump and pg_restore utilities.
  • Executed inside the operating system terminal, not inside the mongosh shell.
  • Exports collections as binary .bson files and index mappings as .metadata.json.
  • The .metadata.json file allows mongorestore to rebuild indexes automatically.
  • Use the --gzip flag to save disk space and reduce network load.
  • Essential for migrating database snapshots between local staging and cloud staging.
Built with LogoFlowershow