This article analyzes frequent timeout alerts caused by the getUiToken API, investigates JVM garbage‑collection pauses linked to excessive PhantomReference objects from MySQL connections, and presents configuration, code, and scheduling solutions that dramatically reduce GC latency and improve service stability.
When a MySQL master‑slave switch makes the old master read-only, Go connection pools keep reusing stale connections and cause prolonged write errors, but by extending the driver to return driver.ErrBadConn on error 1290 or using the DSN flag rejectReadOnly=true, the bad connections are discarded and the error disappears almost instantly.
Jedis uses Apache Commons’ GenericObjectPool to manage Redis connections, with Spring‑Data‑Redis creating a JedisConnectionFactory that holds a configurable pool where borrowObject and returnObject handle idle queues, eviction and validation settings (testOnBorrow, testWhileIdle, etc.) can be tuned, and JMX metrics aid troubleshooting.
An in‑depth analysis of a production incident where HttpClient connections accumulated in CLOSE_WAIT, covering symptom identification, TCP state inspection, root‑cause debugging, HttpClient pool internals, and practical configuration tweaks to prevent similar outages.
After a week of API timeouts in an internal sandbox, we traced the issue to Redis connection pool deadlocks caused by improper Jedis resource handling in Spring, revealing how missing maxWaitMillis settings and unreleased connections can freeze all HTTP threads.
The article investigates a DBCP 1.4 connection‑pool bug that allows the pool to exceed its configured limits, explains how removeAbandoned and socketTimeout interact to produce excess MySQL connections, reproduces the issue with a multithreaded test, and proposes configuration changes to avoid it.
The article explains how reusing HTTP connections with OkHttp—by leveraging persistent HTTP/1.1, multiplexed HTTP/2, and pre‑established sockets—dramatically cuts latency, details the underlying ConnectInterceptor and ConnectionPool implementation, and recommends enlarging the pool and adjusting keep‑alive settings to suit multi‑domain, high‑concurrency apps.
A high‑traffic promotion system suffered massive thread‑pool exhaustion and service outages due to an incorrectly configured HttpClient connection pool, leading to port binding failures, CPU spikes, and instance crashes; the post details the diagnosis, root cause, and concrete mitigation steps.
This article demystifies MyBatis transaction handling by clarifying the true JDBC transaction methods, explaining Spring's propagation concepts, detailing the MyBatis Transaction and TransactionFactory interfaces, showcasing code examples, and exploring special cases such as auto‑commit nuances and connection‑pool interactions.
This guide shows how to create a custom HttpRequestRetryHandler that intelligently retries failed HTTP calls—handling timeouts, connection errors, SSL issues, and idempotent requests—and integrates it into a CloseableHttpClient built with a connection pool and custom request configuration.
The article explains that using the deprecated request.releaseConnection() with Apache HttpClient 4.5.5 can cause a java.net.SocketException: socket closed during high‑frequency API testing, and recommends switching to PoolingHttpClientConnectionManager to manage connections safely.
