企业官网建设流程全解析

如何降低网站相似度,未备案个人网站 如何挣钱,房产信息网510,广州贝勤网络科技有限公司JVM Native Memory Tracking (NMT) 追踪堆外内存泄漏 #x1f4ca; NMT基础概念 什么是NMT#xff1f; Native Memory Tracking是HotSpot JVM内置的追踪工具#xff0c;用于监控JVM内部的内存使用情况。 监控范围 bash 复制 下载 # NMT追踪的内存区域#xff1a; 1. …JVM Native Memory Tracking (NMT) 追踪堆外内存泄漏NMT基础概念什么是NMTNative Memory Tracking是HotSpot JVM内置的追踪工具用于监控JVM内部的内存使用情况。监控范围bash复制下载# NMT追踪的内存区域 1. Java Heap # Java堆 2. Class # 类元数据 3. Thread # 线程栈 4. Code # JIT编译代码 5. GC # 垃圾收集器 6. Compiler # 编译器 7. Internal # 命令行解析、JVMTI等 8. Symbol # 符号表 9. Native Memory Tracking # NMT自身开销 10. Arena Chunk # 分配器竞技场 11. Test # 测试代码 12. Other # 未分类内存NMT启用与配置启动参数配置bash复制下载# 基础启用 java -XX:NativeMemoryTrackingsummary -jar app.jar # 详细模式提供更多细节 java -XX:NativeMemoryTrackingdetail -jar app.jar # 基准模式记录初始内存使用 java -XX:NativeMemoryTrackingsummary -XX:UnlockDiagnosticVMOptions \ -XX:PrintNMTStatistics -jar app.jar运行时监控命令bash复制下载# 1. 获取当前内存使用情况 jcmd pid VM.native_memory summary # 2. 获取详细内存使用情况 jcmd pid VM.native_memory detail # 3. 获取按类别统计 jcmd pid VM.native_memory summary scaleMB # 4. 基线测量记录当前状态作为基准 jcmd pid VM.native_memory baseline # 5. 对比差异与基线对比 jcmd pid VM.native_memory summary.diff # 6. 详细差异对比 jcmd pid VM.native_memory detail.diff # 7. 按类别显示差异 jcmd pid VM.native_memory summary.diff scaleKB常见堆外内存泄漏场景1.Direct ByteBuffer泄漏java复制下载// 常见的DirectByteBuffer泄漏代码 public class DirectMemoryLeak { private ListByteBuffer bufferList new ArrayList(); public void leakMemory() { // 每次分配100MB直接内存 ByteBuffer buffer ByteBuffer.allocateDirect(100 * 1024 * 1024); bufferList.add(buffer); // 忘记清理 } // 正确做法需要显式释放 public void cleanDirectBuffer(ByteBuffer buffer) { if (buffer.isDirect()) { // 需要等GC回收或手动调用Cleaner ((DirectBuffer) buffer).cleaner().clean(); } } }2.MappedByteBuffer泄漏java复制下载public class MappedBufferLeak { private ListMappedByteBuffer mappedBuffers new ArrayList(); public void mapFile(String filePath) throws IOException { RandomAccessFile file new RandomAccessFile(filePath, rw); FileChannel channel file.getChannel(); // 内存映射会占用Native Memory MappedByteBuffer buffer channel.map( FileChannel.MapMode.READ_WRITE, 0, channel.size() ); mappedBuffers.add(buffer); // 忘记关闭channel.close() 和 file.close() } }3.JNI代码内存泄漏java复制下载public class JNIMemoryLeak { static { System.loadLibrary(nativeLib); } // Native方法可能分配内存 public native void allocateNativeMemory(long size); public native void freeNativeMemory(); public void causeLeak() { allocateNativeMemory(1024 * 1024); // 分配1MB // 忘记调用freeNativeMemory() } }4.线程创建过多java复制下载public class ThreadLeak { private ExecutorService executor Executors.newCachedThreadPool(); public void createThreads() { for (int i 0; i 10000; i) { executor.submit(() - { try { Thread.sleep(Long.MAX_VALUE); } catch (InterruptedException e) { e.printStackTrace(); } }); } // 每个线程栈占用1MB默认10k线程占用10GB Native Memory } }NMT输出解析示例输出分析text复制下载Native Memory Tracking: Total: reserved6643041KB, committed397016KB - Java Heap (reserved5070848KB, committed317440KB) (mmap: reserved5070848KB, committed317440KB) - Class (reserved1070513KB, committed12609KB) (classes #1199) (malloc9329KB #1525) (mmap: reserved1067184KB, committed3280KB) - Thread (reserved26546KB, committed26546KB) (thread #26) (stack: reserved26432KB, committed26432KB) (malloc96KB #52) (arena18KB #50) - Code (reserved249632KB, committed2592KB) (malloc32KB #299) (mmap: reserved249600KB, committed2560KB) - GC (reserved47739KB, committed47739KB) (malloc10423KB #117) (mmap: reserved37316KB, committed37316KB) - Compiler (reserved132KB, committed132KB) (malloc1KB #21) (arena131KB #3) - Internal (reserved580KB, committed580KB) (malloc548KB #651) (mmap: reserved32KB, committed32KB) - Symbol (reserved1525KB, committed1525KB) (malloc976KB #111) (arena549KB #1) - Native Memory Tracking (reserved161KB, committed161KB) (malloc99KB #1561) (tracking overhead62KB) - Arena Chunk (reserved175KB, committed175KB) (malloc175KB) - Unknown (reserved33KB, committed33KB) (mmap: reserved33KB, committed33KB)关键指标解读bash复制下载# 重点监控区域 1. Thread区域持续增长 → 线程泄漏 2. Internal区域异常大 → 可能DirectByteBuffer泄漏 3. GC区域过大 → GC配置问题 4. Arena Chunk异常增长 → 内存池泄漏堆外内存泄漏诊断流程步骤1启用NMT监控bash复制下载# 启动应用 java -XX:NativeMemoryTrackingdetail \ -XX:UnlockDiagnosticVMOptions \ -XX:PrintNMTStatistics \ -jar application.jar # 获取进程ID jps -l # 建立基线 jcmd pid VM.native_memory baseline步骤2模拟操作并收集数据bash复制下载# 执行疑似泄漏的操作 curl http://localhost:8080/leaky-endpoint # 等待一段时间 sleep 30 # 获取差异报告 jcmd pid VM.native_memory summary.diff篇幅限制下面就只能给大家展示小册部分内容了。整理了一份核心面试笔记包括了Java面试、Spring、JVM、MyBatis、Redis、MySQL、并发编程、微服务、Linux、Springboot、SpringCloud、MQ、Kafc需要全套面试笔记及答案【点击此处即可/免费获取】​​​步骤3分析泄漏点bash复制下载# 详细差异分析 jcmd pid VM.native_memory detail.diff nmt_diff.txt # 重点关注增长的区域 grep -A 5 Total: nmt_diff.txt grep committed nmt_diff.txt | sort -k2 -nr步骤4使用pmap进一步分析bash复制下载# 查看进程内存映射 pmap -x pid | sort -k2 -nr | head -20 # 查看anon内存段 pmap pid | grep anon | awk {sum$2} END {print sum}高级诊断技巧结合其他工具分析bash复制下载# 1. 使用gdb分析Native Memory gdb -p pid (gdb) malloc_info 0 /tmp/malloc_info.xml # 2. 使用strace追踪系统调用 strace -f -e tracemmap,munmap,brk -p pid 21 | grep -v ENOENT # 3. 使用/proc文件系统 cat /proc/pid/maps | grep heap # 查看堆内存 cat /proc/pid/smaps | grep -A 15 heap # 详细堆信息 # 4. 监控系统内存 watch -n 1 ps aux | grep java | grep -v grep自动化监控脚本bash复制下载#!/bin/bash # monitor_nmt.sh PID$1 INTERVAL10 OUTPUT_DIR./nmt_logs mkdir -p $OUTPUT_DIR while true; do TIMESTAMP$(date %Y%m%d_%H%M%S) # 收集NMT数据 jcmd $PID VM.native_memory detail $OUTPUT_DIR/nmt_detail_$TIMESTAMP.txt jcmd $PID VM.native_memory summary $OUTPUT_DIR/nmt_summary_$TIMESTAMP.txt # 收集系统内存信息 ps -p $PID -o pid,rss,vsz,pcpu,pmem,cmd $OUTPUT_DIR/system_mem_$TIMESTAMP.txt pmap $PID | tail -1 $OUTPUT_DIR/pmap_$TIMESTAMP.txt # 分析增长趋势 if [ -f $OUTPUT_DIR/nmt_summary_previous.txt ]; then diff $OUTPUT_DIR/nmt_summary_previous.txt \ $OUTPUT_DIR/nmt_summary_$TIMESTAMP.txt \ $OUTPUT_DIR/diff_$TIMESTAMP.txt fi cp $OUTPUT_DIR/nmt_summary_$TIMESTAMP.txt \ $OUTPUT_DIR/nmt_summary_previous.txt sleep $INTERVAL done️代码层面的预防措施1.DirectByteBuffer管理类java复制下载public class DirectMemoryManager { private static final ListByteBuffer BUFFERS new ArrayList(); private static final Cleaner CLEANER Cleaner.create(); public static ByteBuffer allocateDirect(long size) { ByteBuffer buffer ByteBuffer.allocateDirect((int) size); BUFFERS.add(buffer); // 注册清理操作 CLEANER.register(buffer, () - { if (buffer.isDirect()) { ((DirectBuffer) buffer).cleaner().clean(); } }); return buffer; } public static void cleanAll() { BUFFERS.forEach(buffer - { if (buffer.isDirect() buffer.capacity() 0) { ((DirectBuffer) buffer).cleaner().clean(); } }); BUFFERS.clear(); } }2.资源关闭模板java复制下载public class ResourceTemplate { public static void withMappedBuffer(String filePath, ConsumerMappedByteBuffer action) { RandomAccessFile file null; FileChannel channel null; try { file new RandomAccessFile(filePath, rw); channel file.getChannel(); MappedByteBuffer buffer channel.map( FileChannel.MapMode.READ_WRITE, 0, channel.size() ); action.accept(buffer); } catch (IOException e) { throw new RuntimeException(e); } finally { try { if (channel ! null) channel.close(); if (file ! null) file.close(); } catch (IOException e) { // 记录日志 } } } }3.线程池监控java复制下载public class MonitoredThreadPool { private final ThreadPoolExecutor executor; private final ScheduledExecutorService monitor; public MonitoredThreadPool() { this.executor new ThreadPoolExecutor( 10, 100, 60L, TimeUnit.SECONDS, new LinkedBlockingQueue(1000) ); // 监控线程池状态 this.monitor Executors.newSingleThreadScheduledExecutor(); this.monitor.scheduleAtFixedRate(() - { System.out.println(Thread pool stats:); System.out.println( Active threads: executor.getActiveCount()); System.out.println( Pool size: executor.getPoolSize()); System.out.println( Queue size: executor.getQueue().size()); // 检查Native Memory使用 try { Runtime runtime Runtime.getRuntime(); Process process runtime.exec(pmap getPid()); // 解析输出... } catch (IOException e) { e.printStackTrace(); } }, 0, 30, TimeUnit.SECONDS); } private String getPid() { return ManagementFactory.getRuntimeMXBean().getName().split()[0]; } }篇幅限制下面就只能给大家展示小册部分内容了。整理了一份核心面试笔记包括了Java面试、Spring、JVM、MyBatis、Redis、MySQL、并发编程、微服务、Linux、Springboot、SpringCloud、MQ、Kafc需要全套面试笔记及答案【点击此处即可/免费获取】​​​监控告警配置Prometheus监控指标yaml复制下载# NMT监控指标采集 - job_name: jvm_nmt static_configs: - targets: [localhost:12345] metrics_path: /nmt/metrics params: pid: [java_pid]Grafana仪表板配置json复制下载{ panels: [ { title: Native Memory Usage, targets: [ { expr: jvm_nmt_memory_reserved{region\Internal\}, legendFormat: Internal Memory }, { expr: jvm_nmt_memory_committed{region\Thread\}, legendFormat: Thread Stack } ] } ] }紧急处理方案内存泄漏紧急处理bash复制下载# 1. 立即收集证据 jcmd pid VM.native_memory detail emergency_nmt_$(date %s).txt jstack pid emergency_threads_$(date %s).txt # 2. 尝试清理DirectBuffer jmap -histo:live pid | grep DirectByteBuffer # 3. 如果可能触发Full GC jcmd pid GC.run # 4. 限制Native Memory使用 # 重启时添加参数 -XX:MaxDirectMemorySize512m # 限制直接内存 -XX:ThreadStackSize256k # 减小线程栈大小 -XX:MetaspaceSize128m # 限制元空间最佳实践总结预防措施定期监控生产环境开启NMT定期收集数据代码审查重点审查DirectByteBuffer、MappedByteBuffer使用资源管理使用try-with-resources或模板方法容量规划合理设置JVM参数限制Native Memory压力测试模拟长时间运行观察Native Memory增长诊断要点基线对比使用baseline和diff功能趋势分析关注committed内存的持续增长区域定位确定是Thread、Internal还是Arena泄漏结合工具NMTpmapjstack综合分析调优建议bash复制下载# 推荐配置 -XX:NativeMemoryTrackingsummary # 生产环境开启 -XX:UnlockDiagnosticVMOptions -XX:PrintNMTStatistics # JVM退出时打印统计 -XX:MaxDirectMemorySize1g # 限制直接内存 -XX:ThreadStackSize512k # 优化线程栈大小 -XX:MetaspaceSize256m # 设置合适的元空间 -XX:MaxMetaspaceSize512m # 限制最大元空间通过系统化的NMT监控和诊断可以有效发现和解决堆外内存泄漏问题保证Java应用的稳定运行。