Delete and TTL Indexes

Introduction

Removing data is as important as inserting it. MongoDB offers deleteOne, deleteMany, and collection-level drop. For temporary data—sessions, OTP codes, raw events—TTL indexes expire documents automatically. This chapter covers safe deletes, soft-delete patterns, and TTL setup.

Prerequisites

• Update Operators and Upsert
• Query Operators

deleteOne and deleteMany

use hello_demo
 
// Remove one matching document
db.posts.deleteOne({ slug: "draft-notes" })
 
// Remove all unpublished drafts
db.posts.deleteMany({ published: false })
 
// DANGEROUS — empty filter deletes everything in collection
// db.posts.deleteMany({})

Return value: deletedCount.

Soft Delete Pattern

Keep rows for audit; hide in queries:

db.posts.updateOne(
  { slug: "mongodb-basics" },
  { $set: { deletedAt: new Date() } }
)
 
// Application queries exclude soft-deleted
db.posts.find({ deletedAt: { $exists: false } })
db.posts.find({ deletedAt: null })

Restore:

db.posts.updateOne(
  { slug: "mongodb-basics" },
  { $unset: { deletedAt: "" } }
)

Relational equivalent: deleted_at column in MySQL UPDATE/DELETE patterns.

findOneAndDelete

Atomic delete and return removed doc:

db.sessions.findOneAndDelete({ token: "abc123" })

Useful for one-time tokens or queue consumption patterns.

drop Collection and Database

// Remove entire collection and its indexes
db.temp_logs.drop()
 
// Remove database — all collections gone
db.getSiblingDB("hello_demo").dropDatabase()

Irreversible on standalone without backup—use mongodump first — Insert and Import Export.

TTL Indexes (Time To Live)

MongoDB deletes documents when indexed date field + expireAfterSeconds elapses.

Session example

db.sessions.createIndex(
  { createdAt: 1 },
  { expireAfterSeconds: 3600 }
)
 
db.sessions.insertOne({
  userId: "user-1",
  token: "secret",
  createdAt: new Date()
})

Background thread removes expired documents—may lag slightly; not a real-time timer per millisecond.

OTP / verification codes

db.verification_codes.createIndex(
  { expiresAt: 1 },
  { expireAfterSeconds: 0 }
)
 
db.verification_codes.insertOne({
  email: "ada@example.com",
  code: "482910",
  expiresAt: new Date(Date.now() + 10 * 60 * 1000)
})

expireAfterSeconds: 0 expires at absolute time in expiresAt.

TTL rules

Compare Redis key TTL — Redis expiration for cache; MongoDB TTL for durable short-lived documents.

When to Use TTL vs Application Delete

Cleanup Before Drop

Dev reset:

db.posts.deleteMany({})
db.posts.dropIndexes()

Production: never dropDatabase without backup and approval.

Common Mistakes

Post-Chapter Checklist

• deleteOne / deleteMany with explicit filters
• Soft delete with deletedAt + query filter
• TTL index on sessions or verification_codes
• You know drop() vs deleteMany

FAQ

Recover deleted documents?

Only from backup (mongorestore) or if soft-deleted.

TTL vs capped collections?

Capped collections are FIFO fixed size—legacy; TTL is more flexible.

Delete performance?

Large deletes may block; batch delete by _id range in maintenance window.

GDPR "right to be forgotten"?

deleteOne({ userId }) across collections or anonymize fields—design in Data modeling.

Redis DEL vs MongoDB delete?

Redis DEL removes cache keys instantly; MongoDB delete removes persisted documents—different layers.