How would you diagnose a slow Spring Boot endpoint in production?

Moves from user-visible latency to route, dependency, database, executor, JVM, and change evidence before selecting a fix.

1Say itAnswer out loud first

2CompareCompare with Core answer

3RefineRefine with Stronger and Advanced

TL;DR

I define the scope first: affected route, percentiles, time window, tenants or inputs, error rate, and whether the regression matches a deployment or traffic change. A representative distributed trace shows time in controller, database, downstream calls, and queueing. I correlate it with structured logs and metrics rather than reading one slow request in isolation.

Full answer below

Practice focus

Latency distributions, traces, structured logs, query and pool metrics, thread dumps, JVM signals, deployment correlation, and mitigation.

Answer first

Answer path

Core answer first, then checkpoints toward a stronger answer.

Start
Core answer
Start here. This is the direct spoken answer to practice first.
Overview
A slow endpoint is a symptom; diagnosis identifies where the request spends time and what changed before tuning anything.
I define the scope first: affected route, percentiles, time window, tenants or inputs, error rate, and whether the regression matches a deployment or traffic change. A representative distributed trace shows time in controller, database, downstream calls, and queueing. I correlate it with structured logs and metrics rather than reading one slow request in isolation.
Checkpoint
Stronger answer
Optional upgrade with examples, trade-offs, and sharper phrasing.
If database spans dominate, I inspect generated SQL, query plans, lock waits, row counts, and Hikari acquisition time. If time is outside spans or tasks wait, I check server threads, executor queues, downstream client pools, and a thread dump. JVM CPU, allocation, garbage collection, and memory pressure can explain broad pauses. I compare fast and slow examples and keep route labels normalized so the telemetry itself is trustworthy.

Review signals

Use these to catch weak phrasing, shallow reasoning, and risky claims.

Common mistakesCandidateCommon mistakesMistakes that weaken the answer or signal shallow understanding.Open details

Details

Looking only at average latency.
Adding a cache before locating the slow boundary.
Restarting before collecting representative evidence.

How would you diagnose a slow Spring Boot endpoint in production?

Practice focus

Answer path

Overview

Review signals

Follow-up prompts

Continue the section

Tech stacks

Skill areas

Suggest an improvement

Found this answer useful?