Question 1

How do you profile a slow Node service?

Accepted Answer

CPU profiles, flamegraphs, heap snapshots, event-loop delay metrics, and APM traces — hypothesize then measure. Avoid premature optimization. Reproduce with load tests. Distinguish JS CPU, GC, I/O wait, and lock time.

Question 2

When do worker_threads help vs hurt?

Accepted Answer

Help CPU-bound parallel work; hurt when serialization/copy overhead dominates or work is tiny. Share ArrayBuffers carefully. Prefer pooling workers. Don't expect magic for I/O-bound tasks.

Question 3

JSON.stringify on huge objects in the request path — issue?

Accepted Answer

CPU cost and event-loop blocking; also memory spikes. Stream responses, paginate, or move heavy serialization off hot paths. Consider faster serializers when measured.

Question 4

Connection pooling best practices for Node DB clients?

Accepted Answer

Size pools to DB limits Ã— instances, set timeouts, and monitor wait queue latency. Too large pools stampede the DB. Too small increases queue time. Align with horizontal pod counts.

Question 5

What Node flags/metrics help GC tuning conversations?

Accepted Answer

heapUsed/heapTotal, gc events (perf_hooks), and understanding young/old space — tune only with evidence. Usually fix allocations first. Increasing heap can mask leaks and raise pause times.

Question 6

Lead: define performance budgets for Node microservices.

Accepted Answer

p50/p99 latency, error rate, event-loop delay, heap growth SLO, and CI load-test gates on critical paths. Assign owners per service. Budget regressions block merges. Pair with capacity planning.

Performance Node.js Interview Questions

Other Node.js interview topics