性能调优命令之jstack

jstack是java虚拟机自带的一种线程堆栈跟踪工具。

/opt/java8/bin/jstack

Usage:
    jstack [-l] <pid>
        (to connect to running process)   #连接活动线程
    jstack -F [-m] [-l] <pid>
        (to connect to a hung process)    #连接阻塞线程
    jstack [-m] [-l] <executable> <core>
        (to connect to a core file)       #连接dump的文件
    jstack [-m] [-l] [server_id@]<remote server IP or hostname>
        (to connect to a remote debug server)     #连接远程服务器

Options:
    -F  to force a thread dump. Use when jstack <pid> does not respond (process is hung)
    -m  to print both java and native frames (mixed mode)
    -l  long listing. Prints additional information about locks
    -h or -help to print this help message

$ jstack 5611
2021-06-06 22:05:43
Full thread dump Java HotSpot(TM) Client VM (25.271-b09 mixed mode):

"RemoteInvocationHandler [#2]" #63304 daemon prio=5 os_prio=0 tid=0xe3f4bc00 nid=0x240 in Object.wait() [0xe3b3e000]
   java.lang.Thread.State: TIMED_WAITING (on object monitor)
	at java.lang.Object.wait(Native Method)
	at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:144)
	- locked <0xea7480c8> (a java.lang.ref.ReferenceQueue$Lock)
	at hudson.remoting.RemoteInvocationHandler$Unexporter.run(RemoteInvocationHandler.java:603)
	at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
	at hudson.remoting.AtmostOneThreadExecutor$Worker.run(AtmostOneThreadExecutor.java:118)
	at java.lang.Thread.run(Thread.java:748)

"Attach Listener" #63199 daemon prio=9 os_prio=0 tid=0xe2a52800 nid=0x7938 waiting on condition [0x00000000]
   java.lang.Thread.State: RUNNABLE

"Ping thread for channel hudson.remoting.Channel@4afd8d:channel" #24 daemon prio=5 os_prio=0 tid=0xe4637000 nid=0x1617 waiting on condition [0xe34b9000]
   java.lang.Thread.State: TIMED_WAITING (sleeping)
	at java.lang.Thread.sleep(Native Method)
	at hudson.remoting.PingThread.run(PingThread.java:95)

"Channel reader thread: channel" #9 prio=5 os_prio=0 tid=0xe3f50800 nid=0x15fa runnable [0xe3aed000]
   java.lang.Thread.State: RUNNABLE
	at java.io.FileInputStream.readBytes(Native Method)
	at java.io.FileInputStream.read(FileInputStream.java:255)
	at java.io.BufferedInputStream.fill(BufferedInputStream.java:246)
	at java.io.BufferedInputStream.read(BufferedInputStream.java:265)
	- locked <0xea4b13b0> (a java.io.BufferedInputStream)
	at hudson.remoting.FlightRecorderInputStream.read(FlightRecorderInputStream.java:92)
	at hudson.remoting.ChunkedInputStream.readHeader(ChunkedInputStream.java:72)
	at hudson.remoting.ChunkedInputStream.readUntilBreak(ChunkedInputStream.java:103)
	at hudson.remoting.ChunkedCommandTransport.readBlock(ChunkedCommandTransport.java:39)
	at hudson.remoting.AbstractSynchronousByteArrayCommandTransport.read(AbstractSynchronousByteArrayCommandTransport.java:34)
	at hudson.remoting.SynchronousCommandTransport$ReaderThread.run(SynchronousCommandTransport.java:63)

"Service Thread" #6 daemon prio=9 os_prio=0 tid=0xf7085400 nid=0x15f2 runnable [0x00000000]
   java.lang.Thread.State: RUNNABLE

"C1 CompilerThread0" #5 daemon prio=9 os_prio=0 tid=0xf7082000 nid=0x15f1 waiting on condition [0x00000000]
   java.lang.Thread.State: RUNNABLE

"Signal Dispatcher" #4 daemon prio=9 os_prio=0 tid=0xf7080800 nid=0x15f0 runnable [0x00000000]
   java.lang.Thread.State: RUNNABLE

"Finalizer" #3 daemon prio=8 os_prio=0 tid=0xf7066400 nid=0x15ef in Object.wait() [0xe45ad000]
   java.lang.Thread.State: WAITING (on object monitor)
	at java.lang.Object.wait(Native Method)
	at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:144)
	- locked <0xea486e40> (a java.lang.ref.ReferenceQueue$Lock)
	at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:165)
	at java.lang.ref.Finalizer$FinalizerThread.run(Finalizer.java:216)

"Reference Handler" #2 daemon prio=10 os_prio=0 tid=0xf7063800 nid=0x15ee in Object.wait() [0xe45fe000]
   java.lang.Thread.State: WAITING (on object monitor)
	at java.lang.Object.wait(Native Method)
	at java.lang.Object.wait(Object.java:502)
	at java.lang.ref.Reference.tryHandlePending(Reference.java:191)
	- locked <0xea486fe0> (a java.lang.ref.Reference$Lock)
	at java.lang.ref.Reference$ReferenceHandler.run(Reference.java:153)

"main" #1 prio=5 os_prio=0 tid=0xf7007400 nid=0x15ec in Object.wait() [0xf717d000]
   java.lang.Thread.State: TIMED_WAITING (on object monitor)
	at java.lang.Object.wait(Native Method)
	at hudson.remoting.Channel.join(Channel.java:1187)
	- locked <0xea747e30> (a hudson.remoting.Channel)
	at hudson.remoting.Launcher.main(Launcher.java:796)
	at hudson.remoting.Launcher.runWithStdinStdout(Launcher.java:718)
	at hudson.remoting.Launcher.run(Launcher.java:398)
	at hudson.remoting.Launcher.main(Launcher.java:298)

"VM Thread" os_prio=0 tid=0xf705e000 nid=0x15ed runnable 

"VM Periodic Task Thread" os_prio=0 tid=0xf7087800 nid=0x15f3 waiting on condition 

JNI global references: 268

线程状态图

jstack统计线程数:
jstack 5611 | grep 'java.lang.Thread.State' | wc -l

举例说明CPU高获取其线程ID然后分析
以我们最近出现的一个实际故障为例，介绍怎么定位和解决这类问题。

根据top命令，发现PID为28555的Java进程占用CPU高达200%，出现故障。
通过ps aux | grep PID命令，可以进一步确定是tomcat进程出现了问题。但是，怎么定位到具体线程或者代码呢？
首先显示线程列表:
ps -mp pid -o THREAD,tid,time
找到了耗时最高的线程28802，占用CPU时间快两个小时了！
其次将需要的线程ID转换为16进制格式：
printf "%x " tid

最后打印线程的堆栈信息：
jstack pid |grep tid -A 30

找到出现问题的代码了！
现在来分析下具体的代码：ShortSocketIO.readBytes(ShortSocketIO.java:106)
ShortSocketIO是应用封装的一个用短连接Socket通信的工具类。readBytes函数的代码如下：

public byte[] readBytes(int length) throws IOException {
    if ((this.socket == null) || (!this.socket.isConnected())) {
        throw new IOException("++++ attempting to read from closed socket");
    }
    byte[] result = null;
    ByteArrayOutputStream bos = new ByteArrayOutputStream();
    if (this.recIndex >= length) {
           bos.write(this.recBuf, 0, length);
           byte[] newBuf = new byte[this.recBufSize];
           if (this.recIndex > length) {
               System.arraycopy(this.recBuf, length, newBuf, 0, this.recIndex - length);
           }
           this.recBuf = newBuf;
           this.recIndex -= length;
    } else {
           int totalread = length;
           if (this.recIndex > 0) {
                totalread -= this.recIndex;
                bos.write(this.recBuf, 0, this.recIndex);
                this.recBuf = new byte[this.recBufSize];
                this.recIndex = 0;
    }
    int readCount = 0;
  **  while (totalread > 0) {
         if ((readCount = this.in.read(this.recBuf)) > 0) **{
                if (totalread > readCount) {
                      bos.write(this.recBuf, 0, readCount);
                      this.recBuf = new byte[this.recBufSize];
                      this.recIndex = 0;
               } else {
                     bos.write(this.recBuf, 0, totalread);
                     byte[] newBuf = new byte[this.recBufSize];
                     System.arraycopy(this.recBuf, totalread, newBuf, 0, readCount - totalread);
                     this.recBuf = newBuf;
                     this.recIndex = (readCount - totalread);
             }
             totalread -= readCount;
        }
   }
}

问题就出在加粗的代码部分。如果this.in.read()返回的数据小于等于0时，循环就一直进行下去了。而这种情况在网络拥塞的时候是可能发生的。
至于具体怎么修改就看业务逻辑应该怎么对待这种特殊情况了。

最后，总结下排查CPU故障的方法和技巧有哪些：
1、top命令：Linux命令。可以查看实时的CPU使用情况。也可以查看最近一段时间的CPU使用情况。
2、PS命令：Linux命令。强大的进程状态监控命令。可以查看进程以及进程中线程的当前CPU使用情况。属于当前状态的采样数据。
3、jstack：Java提供的命令。可以查看某个进程的当前线程栈运行情况。根据这个命令的输出可以定位某个进程的所有线程的当前运行状态、运行代码，以及是否死锁等等。
4、pstack：Linux命令。可以查看某个进程的当前线程栈运行情况。

Java 系统性能分析 命令

1. cpu分析
   top ， pidstat(sysstat)
   pid -p PID -t 1 10
   vmstat 1 CPU上下文切换、运行队列、利用率
   ps Hh -eo tid
   pcpu 查看具体线程的CPU消耗
   sar 来查看一定世界范围内以及历史的cpu消耗情况信息

查看java线程信息
jstack pid | grep 'nid=0x9999'

2. cs sy消耗比较高
   上下文切换性能偏高， jstack -l pid, 查看on object monitor

3. io消耗
   pidstat -d -t -p pid 1 100
   iostat

4. 网络io消耗
   cat /proc/interruptes
   sar -n FULL 1 2