Wrote a few small pieces of code, broke the system, and was blasted by the boss

Java The program is based on GC Of , At the beginning of startup , I applied for a sufficient amount of memory pool , Plus JIT And the real-time optimization of the compiler , It's no faster than using C++ Slow language writing .Java Language is both reflective and observable , Plus JFR This artifact , When a problem occurs, it is easier to find its root than binary files .

Recently in to see RCA（Root Cause Analysis） Things that are , Accidentally found out yCrash Such a thing . It's a few small pieces of problem code, which is very typical , We can take a little look at , Let's see Java Several common crash scenarios for applications .

1. Heap space overflow

OOM It's usually caused by a memory leak , Performance in GC In the Journal , In general, it's GC It's getting longer , And the effect of each recovery is very general .GC after , The actual usage of heap memory is on the rise .

The following code is an endless loop , Keep going to HashMap Richards data , because myMap Belong to GCRoots, Never released , So the end result is OOM.

import java.util.HashMap;
public class OOMDemo {
   static HashMap<Object, Object> myMap = new HashMap<>();
   public static void start() throws Exception { 
      while (true) {
         myMap.put("key" + counter, "Large stringgggggggggggggggggggggggggggg"
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + "ggggggggggggggggggggggggggggggggggggggggggggggggggggg" 
               + counter);         
         ++counter;
      }
   }   
}

2. Memory leak

Memory leaks and memory overflows are the same thing , The difference is its meaning .

Memory overflow may be due to excessive requests , Or the consequences of real business needs , The memory overflow is unknown 、 Beyond expectations OOM situation .

We can use the same code above to achieve this .

In reality , Memory leaks are usually very hidden , Need help Mat Wait for tools to find the root cause .jmap、pmap And so on are common tools .

such as , If you forget to rewrite the object hashCode and equals Method , There will be a memory leak .

//leak example : created by xjjdog 2022
import java.util.HashMap;
import java.util.Map;
public class HashMapLeakDemo {
    public static class Key {
        String title;
        public Key(String title) {
            this.title = title;
        }
    }
    public static void main(String[] args) {
        Map<Key, Integer> map = new HashMap<>();
        map.put(new Key("1"), 1);
        map.put(new Key("2"), 2);
        map.put(new Key("3"), 2);
        Integer integer = map.get(new Key("2"));
        System.out.println(integer);
    }
}

3.CPU soaring

Direct a dead cycle , You can put the CPU Dry to death .

public class CPUSpikeDemo {
  public static void start() {
    new CPUSpikerThread().start();
    new CPUSpikerThread().start();
    new CPUSpikerThread().start();
    new CPUSpikerThread().start();
    new CPUSpikerThread().start();
    new CPUSpikerThread().start();
    System.out.println("6 threads launched!");
  }
}
public class CPUSpikerThread extends Thread {
  @Override
  public void run() {
    while (true) {
      // Just looping infinitely
    }
  }
}

To get the problem code, you can usually use the following method .

（1） Use top command , Found using CPU Most of a process , Record its pid. Use Shift + P Shortcut keys can be pressed CPU The usage rate of the .（2） Again using top command , Add -H Parameters , View the... Used in a process CPU The largest number of threads , Record the thread's ID.（3） Use printf function , Will be decimal tid Convert to hex .（4） Use jstack command , see Java Thread stack of process .（5） Use less Command to view the generated file , And find the hexadecimal conversion just now tid, Find the thread context where the problem occurred .

4. Thread leak

Thread resources are expensive . If you keep creating threads , System resources will soon be exhausted . The following code keeps creating threads , If there is more pressure to request at the same time , Most can kill the host .

public class ThreadLeakDemo {
   public static void start() {
      while (true) {
         new ForeverThread().start();
      }
   }
}
public class ForeverThread extends Thread {
   @Override
   public void run() {
      // Put the thread to sleep forever, so they don't die.
      while (true) {
         try {
            // Sleeping for 10 minutes repeatedly
            Thread.sleep(10 * 60 * 1000);
         } catch (Exception e) {}
      }
   }
}

This is violence , This is similar to creating a thread for each request , Or the consequences of creating a thread pool are the same .xjjdog There are also two related thread leakage articles .

• Strongly opposed to the use of Spring Encapsulated multithreaded class ！
• Fatal failure ！ Blew up investors ！

java.lang.OutOfMemoryError: unable to create new native thread

5. Deadlock

Deadlock codes generally do not occur , But once it happens, it's still very serious , Related businesses may not be able to run .

public class DeadLockDemo {
   public static void start() {
      new ThreadA().start();
      new ThreadB().start();
   }
}

public class ThreadA extends Thread {
    @Override 
    public void run() {
        CoolObject.method1();
    }
}

public class ThreadB extends Thread {
    @Override 
     public void run() {
          HotObject.method2();
     } 
}

public class CoolObject {
    public static synchronized void method1() {
       try {
     // Sleep for 10 seconds
     Thread.sleep(10 * 1000);
          } catch (Exception e) {}
          HotObject.method2();
     }
}
      
public class HotObject {
   public static synchronized void method2() {
       try {
          // Sleep for 10 seconds
          Thread.sleep(10 * 1000);
       } catch (Exception e) {}
       CoolObject.method1();
   } 
}

Deadlock is a serious situation ,jstack Will prompt with obvious information . Of course , About threads dump, There are also some online analysis tools available . such as fastthread, But you also need to understand the meaning of these situations .

 

6. Stack overflow

Stack overflow does not cause JVM Process death , harm “ Relatively small ”. The following is a simple code to simulate stack overflow , Just call recursively .

public class StackOverflowDemo {
   public void start() {
      start();
   }
}

adopt -Xss Parameter can set the size of the virtual machine stack . For example, the following command is to set the stack size to 128K.

-Xss128K

If this happens frequently in your application , Try increasing this value . But it is usually caused by program errors , You'd better check your code .

7.Blocked Threads

BLOCKED Is a more serious thread state , When the processing time of the back-end service is very long , The requested thread will enter the waiting state . At this time through jstack To get the stack , You will find that the thread is blocked . It blocks the acquisition of locks （wating to lock）

public class BlockedAppDemo {
   public static void start() {
      for (int counter = 0; counter < 10; ++counter) {
      // Launch 10 threads.
      new AppThread().start();
      }
   }
}
public class AppThread extends Thread {
   @Override
   public void run() {
      AppObject.getSomething();
   }
}
public class AppObject {
   public static synchronized void getSomething() {
      while (true) {
         try {
         Thread.sleep(10 * 60 * 1000);
  } catch (Exception e) {}
      }
  }
}

Once this happens frequently , It proves that your program is too slow . If CPU There is a surplus of resources , You can try to increase the number of threads requested , such as tomcat Is the maximum number of threads .

End

That's what it's all about Java Analysis of several small codes of common faults , Most of the faults can't escape these scenarios . Troubleshooting is usually very energy-consuming , And you have to have access to the cable . How to make some useful tools , Mask these complexities behind , That's what we want .