A detailed post‑mortem of a three‑time service outage reveals how hidden bugs, frequent FullGC, a saturated connection pool, and an unindexed slow SQL query crippled a Spring Boot backend, and shows the step‑by‑step troubleshooting, temporary fixes, and lasting improvements applied.
The article describes a high‑traffic core application with a MySQL master‑slave setup facing high resource usage; it details analysis of hardware, disk, table space, QPS, slow SQL, and presents governance actions including data archiving, moving reads to replicas, a MyBatis interceptor, and monitoring scripts, achieving significant performance improvements.
During a Monday incident where the backend service became unavailable three times, the author details step-by-step investigations—checking frontend responses, container status, JVM and thread pool metrics, and database connection pool usage—ultimately identifying a slow, unindexed SQL query that exhausted connections, and describes the remediation and lessons learned.
This article details a multi‑phase slow‑SQL governance initiative that defines clear goals, metrics, and operational steps—from data preparation and ticket flow to trend analysis and process tracking—culminating in measurable improvements in system stability and performance.
A detailed postmortem shows how a poorly optimized SQL query, triggered by a cache‑removal update, saturated MySQL's CPU and connection pool, causing cascading failures across a sales‑critical system despite low overall traffic.
During nightly peaks, a social‑e‑commerce startup experienced hour‑long service outages due to MySQL timeouts; by analyzing traffic spikes, CPU usage, and slow‑SQL logs, the team identified un‑cached ranking queries and a 20‑minute cache refresh bottleneck, then implemented targeted caching, monitoring scripts, and fallback static pages to eliminate the issue.
This article shares practical collaboration strategies through real‑world cases, explaining why framing proposals around others' benefits, identifying the right partners, and aligning mutual goals can turn complex technical challenges—like slow‑SQL governance and barcode‑scale integration—into successful joint projects.
This article presents a detailed investigation of intermittent slow SQL queries observed in Meituan‑Dianping’s services, identifies the MySQL 5.6 thread‑pool as the primary cause, and demonstrates how adjusting thread‑pool parameters dramatically reduced slow‑query counts and improved overall service availability.
