接上篇文章, 拿SpringBoot举个例
1.1 默认线程池的隐患
Spring Boot的@Async
默认使用SimpleAsyncTaskExecutor
(无复用线程),频繁创建/销毁线程易引发性能问题。
1.2 自定义线程池配置
@Configuration
@EnableAsync
public class AsyncConfig implements AsyncConfigurer {
@Override
public Executor getAsyncExecutor() {
ThreadPoolTaskExecutor executor = new ThreadPoolTaskExecutor();
executor.setCorePoolSize(10); // 核心线程数=CPU核心数×2
executor.setMaxPoolSize(20); // 突发流量缓冲
executor.setQueueCapacity(100); // 根据业务容忍延迟调整
executor.setThreadNamePrefix("Async-");
executor.setRejectedExecutionHandler(new ThreadPoolExecutor.CallerRunsPolicy());
executor.initialize();
return executor;
}
}
// 使用示例
@Service
public class ReportService {
@Async // 指定使用自定义线程池
public CompletableFuture<Report> generateReportAsync(Long id) {
// 模拟耗时操作
Thread.sleep(2000);
return CompletableFuture.completedFuture(new Report(id, "Done"));
}
}
1.3 线程池监控(Micrometer + Prometheus)
# application.yml
management:
endpoints:
web:
exposure:
include: "metrics,prometheus"
metrics:
tags:
application: ${spring.application.name}
@Bean
public MeterBinder threadPoolMetrics(ThreadPoolTaskExecutor executor) {
return registry -> {
Gauge.builder("thread.pool.active", executor, ThreadPoolTaskExecutor::getActiveCount)
.description("当前活跃线程数")
.register(registry);
Gauge.builder("thread.pool.queue.size", executor, e -> e.getThreadPoolExecutor().getQueue().size())
.description("任务队列长度")
.register(registry);
};
}
通过http://localhost:8080/actuator/prometheus
可获取实时指标。
二、Spring Boot内存泄漏排查:一个真实OOM案例
2.1 故障现象
应用运行24小时后出现
java.lang.OutOfMemoryError: Java heap space
GC日志显示老年代占用持续增长
2.2 诊断步骤
步骤1:生成堆转储文件
# 在应用启动命令中添加
-XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/tmp/heapdump.hprof
步骤2:使用MAT分析
打开
heapdump.hprof
,选择Dominator Tree发现
ConcurrentHashMap$Node[]
占用80%内存查看引用链,定位到缓存工具类未清理过期数据
步骤3:代码修复
// 错误代码:静态Map无限增长
public class CacheManager {
private static Map<String, Object> cache = new ConcurrentHashMap<>();
public static void put(String key, Object value) {
cache.put(key, value);
}
}
// 修复:引入Guava Cache自动过期
public class CacheManager {
private static LoadingCache<String, Object> cache = CacheBuilder.newBuilder()
.maximumSize(1000)
.expireAfterWrite(10, TimeUnit.MINUTES)
.build(new CacheLoader<>() {
@Override
public Object load(String key) {
return loadFromDB(key);
}
});
}
三、Spring Data JPA连接池优化(HikariCP实战)
3.1 默认配置风险
Spring Boot默认使用HikariCP,但以下参数需针对性调整:
spring:
datasource:
hikari:
maximum-pool-size: 20 # 默认10,根据DB并发能力调整
connection-timeout: 3000 # 获取连接超时时间(ms)
idle-timeout: 600000 # 空闲连接存活时间(默认10分钟)
max-lifetime: 1800000 # 连接最大生命周期(默认30分钟)
leak-detection-threshold: 5000 # 连接泄漏检测阈值(生产环境建议开启)
3.2 监控集成
@Bean
public MeterBinder hikariMetrics(HikariDataSource dataSource) {
return registry -> {
HikariPoolMXBean pool = dataSource.getHikariPoolMXBean();
Gauge.builder("db.pool.active", pool::getActiveConnections)
.register(registry);
Gauge.builder("db.pool.idle", pool::getIdleConnections)
.register(registry);
Gauge.builder("db.pool.total", pool::getTotalConnections)
.register(registry);
};
}
四、生产级Spring Boot JVM参数模板
4.1 基础参数(JDK11+)
java -jar your-app.jar \
-Xms2g -Xmx2g # 堆内存固定,避免动态调整开销 \
-XX:MaxMetaspaceSize=256m # 防止元空间膨胀 \
-XX:+UseG1GC # 低延迟垃圾回收器 \
-XX:MaxGCPauseMillis=200 # 目标最大停顿时间 \
-Xlog:gc*,gc+heap=debug:file=gc.log:time,uptime:filecount=5,filesize=100m \
-Dspring.profiles.active=prod
4.2 容器环境适配(Dfile.encoding警告修复)
FROM eclipse-temurin:17-jdk
ENV LANG C.UTF-8
ENV JAVA_OPTS="-Dfile.encoding=UTF-8"
五、实战:利用Arthas在线诊断Spring Boot应用
5.1 安装与附加进程
curl -O https://arthas.aliyun.com/arthas-boot.jar
java -jar arthas-boot.jar # 选择目标进程
5.2 常用命令
# 1. 查看实时线程状态
thread -n 3 # 显示CPU占用最高的3个线程
# 2. 监控方法调用耗时
watch com.example.service.*Service * '{params, returnObj}' -x 3
# 3. 动态修改日志级别(无需重启)
logger --name ROOT --level debug