Medium

Nithidol Vacharotayan

In

Dev Genius

by

Nithidol Vacharotayan

·

Oct 6

Our Redis Lock Expired. The Report Job Kept Running.

Two Spring Boot pods processed the same report, exposing the gap between acquiring a lease and controlling work after it expires.

Our Redis Lock Expired. The Report Job Kept Running.

In

Dev Genius

by

Nithidol Vacharotayan

·

Sep 29

When a CSV Download Controls Your PostgreSQL Transaction

A proposed Spring Boot reproduction examines how streaming can reduce heap usage while tying database resource lifetimes to a slow client.

When a CSV Download Controls Your PostgreSQL Transaction

In

Stackademic

by

Nithidol Vacharotayan

·

Sep 24

Spring Boot Saved the Order. Kafka Never Received the Event.

How a PostgreSQL–Kafka dual write created two versions of success — and why a transactional outbox made the failure recoverable.

Spring Boot Saved the Order. Kafka Never Received the Event.

In

Stackademic

by

Nithidol Vacharotayan

·

Sep 15

Our Load Test Passed Because It Never Triggered Retries

How a Spring Boot search API multiplied 300 customer requests into nearly 900 downstream calls — and exposed a missing failure mode in our…

Our Load Test Passed Because It Never Triggered Retries

In

Stackademic

by

Nithidol Vacharotayan

·

Sep 8

The 20 MB File That Changed Our Spring Boot API

Embedding file content in JSON turned a modest upload into unpredictable heap pressure — and showed us that content type is also a resource…

The 20 MB File That Changed Our Spring Boot API

In

Stackademic

by

Nithidol Vacharotayan

·

Sep 2

PostgreSQL Used the Index. The Query Still Read 8.7 Million Rows.

Why an Index Scan gave us false confidence—and how earlier filtering reduced the work before an expensive reporting join.

PostgreSQL Used the Index. The Query Still Read 8.7 Million Rows.

In

Stackademic

by

Nithidol Vacharotayan

·

Aug 26

Our Server Was Healthy. Our Connection Pool Was Exhausted.

How Oracle connection occupancy made a responsive Spring Boot service unusable — and why the fix required both query efficiency and…

Our Server Was Healthy. Our Connection Pool Was Exhausted.

In

Stackademic

by

Nithidol Vacharotayan

·

Aug 20

PostgreSQL Was Fast. Our Spring Boot API Still Ran Out of Memory!

How an unpaginated Hibernate object graph turned a report preview into a full-report workload inside one HTTP request.

PostgreSQL Was Fast. Our Spring Boot API Still Ran Out of Memory!

In

Stackademic

by

Nithidol Vacharotayan

·

Aug 12

Quartz Ran Successfully. The Report Was Still Wrong.

An implicit server timezone shifted both a Spring Boot job and the database window it was supposed to report.

Quartz Ran Successfully. The Report Was Still Wrong.

In

Stackademic

by

Nithidol Vacharotayan

·

Aug 5

Why Our Spring Boot Health Check Lied During an Outage

Every pod remained ready while all 200 checkout threads waited on the same dependency.

Why Our Spring Boot Health Check Lied During an Outage

Nithidol Vacharotayan

Nithidol Vacharotayan

I write about production failures in Java backend systems: Spring Boot, Kafka, Redis, PostgreSQL, and the trade-offs that only show up after launch.

Following

Help

Status

About

Careers

Press

Blog

Store

Privacy

Rules

Terms

Text to speech