MONGODB · HARD CODEX

MongoDB Interview Questions — Hard

Q: Explain WiredTiger storage engine internals, specifically MVCC, page structure, checkpointing, and reconciliation.

Multi-Version Concurrency Control (MVCC): WiredTiger handles concurrent transactions by keeping multiple versions of modified documents. Readers get a point-in-time snapshot, avoiding read locks while

Q: What is read concern `snapshot` and how does it guarantee read isolation in multi-document transactions?

Concept: Only available within multi-document sessions/transactions. Guarantee: It provides strict snapshot isolation. The query engine reads data from a synchronized, point-in-time snapshot of the da

Q: How does the election protocol version 1 (pv1) prevent split-brain scenarios in a distributed replica set?

Split-Brain: A situation where network partitions cause two sub-groups of a replica set to both think they have an active primary, leading to divergent write logs. PV1 Solution: Uses strict raft-like

Q: Explain the step-down process of a Primary node and how to configure priority and votes parameters.

Step-Down: An administrative command (replSetStepDown) or automatic behavior where the active primary relinquishes its role and becomes a secondary. Trigger: Happens during maintenance, or when the pr

Q: How does MongoDB handle node synchronization from scratch (Initial Sync)?

Phase 1: Metadata & Cloning: The syncing node creates database structures, drops existing databases, clones all collections and indexes from a sync source node, while concurrently logging any new oplo

Deep runtime internals, memory models, distributed design, concurrency failure modes, and architectural decisions.

50 Theory Questions7 Implementation Folios5 Free Model Answers

THEORY QUESTIONS & SOLUTIONSShowing 5 of 50 questions

Q1Explain WiredTiger storage engine internals, specifically MVCC, page structure, checkpointing, and reconciliation.

Multi-Version Concurrency Control (MVCC): WiredTiger handles concurrent transactions by keeping multiple versions of modified documents. Readers get a point-in-time snapshot, avoiding read locks while writes occur concurrently.
Page Structure: Data is organized in B-Tree pages (internal and leaf pages) in memory and on disk. Modifications are stored as in-memory "update lists" appended to the original on-disk page layout.
Checkpointing: Every 60 seconds (or when journal writes reach 2GB), WiredTiger flushes all dirty in-memory pages to disk, creating a crash-consistent checkpoint.
Reconciliation: During checkpointing, the engine combines the in-memory update lists with the old on-disk page format, compiles them into a new layout, compresses it, and writes clean, new pages to disk.

Q2What is read concern `snapshot` and how does it guarantee read isolation in multi-document transactions?

Concept: Only available within multi-document sessions/transactions.
Guarantee: It provides strict snapshot isolation. The query engine reads data from a synchronized, point-in-time snapshot of the database across all replica set nodes.
Mechanism: Ensures that all operations inside the transaction observe a single, consistent state of the database. Even if other clients commit transactions concurrently, the transaction's view remains completely isolated and unchanged.

Q3How does the election protocol version 1 (pv1) prevent split-brain scenarios in a distributed replica set?

Split-Brain: A situation where network partitions cause two sub-groups of a replica set to both think they have an active primary, leading to divergent write logs.
PV1 Solution: Uses strict raft-like consensus. A candidate node must receive a majority vote of all voting nodes (e.g., at least 2 out of 3, or 3 out of 5) to become primary.
Double Voting Prevention: A voter can only vote once per election term (using term increment counters). If a network partition occurs, only the partition containing a true majority can elect a primary; the minority partition is physically blocked from electing a primary.

Q4Explain the step-down process of a Primary node and how to configure priority and votes parameters.

Step-Down: An administrative command (replSetStepDown) or automatic behavior where the active primary relinquishes its role and becomes a secondary.
Trigger: Happens during maintenance, or when the primary detects that it has lost connectivity to a majority of the replica set nodes.
Configuration:
- priority: A number (0 to 1000) indicating preference for primary. Secondary nodes with priority: 0 can never become primary.
- votes: A boolean (0 or 1) indicating voting rights. Up to 7 members in a replica set can have votes.

Q5How does MongoDB handle node synchronization from scratch (Initial Sync)?

Phase 1: Metadata & Cloning: The syncing node creates database structures, drops existing databases, clones all collections and indexes from a sync source node, while concurrently logging any new oplog writes occurring on the sync source.
Phase 2: Oplog Application: The syncing node applies the recorded oplog writes to bring its local dataset up to speed.
Pitfalls: If the sync source's oplog rolls over (overwrites older records) before Phase 1 completes, the sync fails, requiring a complete restart.

Unlock the remaining 52 MongoDB (Hard) questions

You've completed the 5 free sample questions. Get unrestricted lifetime access to every question, model answer, implementation challenge, and all 27+ technologies for a single payment.

Unlock Everything — Already Enrolled? Sign In

One-time settlement () · Lifetime access · Zero subscription

CROSS-DIFFICULTY NAVIGATION

Continue Preparing MongoDB

EASY LEVEL

Foundations & Core Concepts

MEDIUM LEVEL

Mechanisms & Practical Patterns