jstack多线程代码调试

1 linux下查询消耗cpu线程排序

指令： top -H -p pid

  [root@outmanage bin]# top -H -p 996

top - 14:32:51 up 1 day, 23:45,  2 users,  load average: 1.00, 0.97, 0.95
Tasks: 183 total,   1 running, 182 sleeping,   0 stopped,   0 zombie
Cpu(s): 34.0%us,  0.6%sy,  0.0%ni, 65.4%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   7927952k total,  7095808k used,   832144k free,   331832k buffers
Swap:  4194300k total,    15488k used,  4178812k free,  2354152k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND                                                                                                    
 1304 root      20   0 5804m 1.2g  19m R 100.1 15.6 145:22.24 OBMServer                                                                                                 
  996 root      20   0 5804m 1.2g  19m S  0.0 15.6   0:00.00 OBMServer                                                                                                  
  997 root      20   0 5804m 1.2g  19m S  0.0 15.6   0:06.05 OBMServer                                                                                                  
  998 root      20   0 5804m 1.2g  19m S  0.0 15.6   0:00.41 OBMServer                                                                                                  
  999 root      20   0 5804m 1.2g  19m S  0.0 15.6   0:00.34 OBMServer                                                                                                  
 1000 root      20   0 5804m 1.2g  19m S  0.0 15.6   0:00.30 OBMServer

2 进制转换

cpu消耗最高1304，十进制转换十六进制为 518

3 线程查询

指令：jstack server_pid |grep thread_16_pid

 [root@outmanage bin]# jstack 996|grep 518
"MailSendThread" #156 prio=5 os_prio=0 tid=0x00007f3dc15d6800 nid=0x518 runnable [0x00007f3d84a12000]

4 代码修改

原因：消费线程缺少通信机制，一直runnable，导致cpu消耗高。
解决：容器改为concurrent包下LinkedBlockingQueue，LinkedBlockingQueue是线程安全的，存和取采用独立的锁，可并行操作队列中的数据，还解决了线程通信问题。

package mailsend;

import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import java.util.concurrent.LinkedBlockingQueue;




public class MailContext {
	//private List<Map> list = new ArrayList<Map>();
	private static MailContext instance = null;
	private LinkedBlockingQueue<Map> queue=new LinkedBlockingQueue(200);
	private MailContext(){}
	
	public synchronized static MailContext getInstance() {
			if(instance == null){//懒汉式 
				instance = new MailContext();
			}
		return instance;
	}
	public synchronized LinkedBlockingQueue<Map> getQueue() {
		System.out.println("MailSendContext:"+queue.size());
		return queue;
	}
//	public synchronized Map getMap() {
//		if (list.size()>0) {
//			System.out.println("MailContext:"+list.size());
//    		return list.remove(0);
//    	}else{
//    		return null;
//    	}
//	}

	
//	public synchronized void setList(Map map) {
//		list.add(map);
//		
//	}

	
}

5 测试

修改后队列为空会阻塞线程，减少性能消耗。

"MailSendThread" #78 prio=5 os_prio=0 tid=0x000000001dc77000 nid=0x5208 waiting on condition [0x000000002ad0e000]
   java.lang.Thread.State: WAITING (parking)
        at sun.misc.Unsafe.park(Native Method)
        - parking to wait for  <0x00000000ed529ba0> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject)
        at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
        at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.await(AbstractQueuedSynchronizer.java:2039)
        at java.util.concurrent.LinkedBlockingQueue.take(LinkedBlockingQueue.java:442)
        at com.ultrapower.nettech.obm.server.mailsend.MailSendThread.run(MailSendThread.java:26)