JVM tuning V: JVM tuning tools and tuning practice

jvm Tuning tools and tuning practices

jvm Built in common commands

JPS

jps Is used to view the hotspot Process of virtual machine id. When... Is not specified hostid when , The default view is native jvm process id

-l: Output the complete jar name
-v: Output starts the specified jvm Parameters

Jmap

This command can be used to view memory information , The number of instances and the size of occupied memory

jmap -histo 25358 # View instances generated in history
jmap -histo:live 25358 # View the currently surviving instances , During execution, a... May be triggered full gc

If there is a large amount of information, you can output it to a file to view jmap -histo 25358 > ./info.txt、less info.txt

num： Serial number
instances： Number of instances
bytes： Occupied space size
class name： Class name ,[C is a char[],[S is a short[],[I is a int[],[B is a byte[],[[I is a int[][]

View heap information
jmap -heap 25358

Export heap information
jmap -dump:format=b,file=bx.hprof 25358 Export heap information , The exported information file can be put into the visualization tool for analysis , for example jvisualvm

When setting memory overflow, export the heap information at that time
-XX:+HeapDumpOnOutOfMemoryError
-XX:HeapDumpPath=./ （ route ）

public class OOMTest {
    
   // JVM Set up  
   // -Xms10M -Xmx10M -XX:+PrintGCDetails -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=D:\jvm.dump 
   public static void main(String[] args) {
    
      List<Object> list = new ArrayList<>();
      int i = 0;
      int j = 0;
      while (true) {
    
         list.add(new User(i++, UUID.randomUUID().toString()));
         new User(j--, UUID.randomUUID().toString());
      }
   }
}

If you want to know about memory usage , You can use this tool for analysis

Jstack

use jstack Add process id Look for deadlocks , See the following example

public class DeadLockTest {
    
	private static Object lock1 = new Object();
	private static Object lock2 = new Object();
	public static void main(String[] args) {
    
		new Thread(() -> {
    
			synchronized (lock1) {
    
				try {
    
					System.out.println("thread1 begin");
					Thread.sleep(5000);
				} catch (InterruptedException e) {
    
				}
				synchronized (lock2) {
    
					System.out.println("thread1 end");
				}
			}
		}).start();
		new Thread(() -> {
    
			synchronized (lock2) {
    
				try {
    
					System.out.println("thread2 begin");
					Thread.sleep(5000);
				} catch (InterruptedException e) {
    
				}
				synchronized (lock1) {
    
					System.out.println("thread2 end");
				}
			}
		}).start();

		System.out.println("main thread end");
	}
}

jstack 22940

“Thread-1” The thread of
prio=5 priority =5
tid=0x00xxxxxxx Threads id
nid=0xxxxx The local thread ID corresponding to the thread nid
java.lang.Thread.State: BLOCKED Thread state

jstack Find out the occupation cpu The highest thread stack information

public class Math {
    
	public int compute() {
      // A method corresponds to a stack frame memory area 
		int a = 1;
		int b = 2;
		int c = (a + b) * 10;
		return c;
	}
	public static void main(String[] args) {
    
		Math math = new Math();
		while (true){
    
			math.compute();
		}
	}

stay Linux Pass through javac Math.java Compile , adopt java Math Run , Pay attention to the package name , It is better to remove the package name , Otherwise, the main class will not be found .
Use top Order to see Linux The state of , notice CPU100%.

top -p 94958 View the status of the specified process

Press H, Get the memory of each thread

Find memory and cpu The highest thread occupied pid, such as 94959
Convert to hexadecimal to get 0x172ef, This is thread id The hexadecimal representation of
perform jstack 94958|grep -A 10 0x172ef, Get thread stack information in 172ef This thread is behind the line 10 That's ok , You can see from the stack what causes cpu Skyrocketing call method

jvisualvm Monitoring tools

Use jvisualvm When detecting the remote server , Additional parameters are required to start the service .

java -Dcom.sun.management.jmxremote.port=8888 -Djava.rmi.server.hostname=192.168.10.111 -Dcom.sun.management.jmxremote.ssl=false -Dcom.sun.management.jmxremote.authenticate=false -jar xxx.jar

function jvisualvm

Jinfo

See what's running Java Extension parameters for the application
see jvm Parameters of operation
jinfo -flags 100054

see java system parameter

Jstat

jstat Command to view the usage of each part of heap memory , And the number of loaded classes . The format of the command is as follows ：
jstat [- Command options ] [vmid] [ Time interval between ( millisecond )] [ Query times ]
Be careful ： The use of jdk The version is jdk8
Garbage collection statistics
jstat -gc pid The most commonly used , Can evaluate program memory usage and GC The overall situation of pressure

S0C: survivor0 District Total size
S0U: survivor0 District Size has been used
EC：Eden The total size of
EU：Eden The size that has been used
OC： The total size of the old age
OU： The old age has used the size
MC： Meta space ( Method area ) The total size of
MU： Meta space ( Method area ) The size that has been used
CCSC: Compress class space size
CCSU: Compressed class space usage size
YGC： The project has been started up to now young GC frequency Above, 69 Time
YGCT： This happened 69 Time spent in total 0.959 second
FGC： The project has been started up to now full Gc frequency Above, 5 Time
FGCT： This happened 5 A total of 1.251 second
GCT： in total GC Time for 2.210

Heap memory statistics
jstat -gccapacity 109691

NGCMN： The minimum capacity of the new generation
NGCMX： The largest capacity of the new generation
NGC： Current Cenozoic capacity
S0C： The size of the first surviving area
S1C： The size of the second surviving area
EC： The size of Eden Park
OGCMN： The smallest capacity in the old days
OGCMX： The biggest capacity in the old days
OGC： The current size of the elderly
OC: The current size of the elderly
MCMN: Minimum metadata capacity
MCMX： Maximum metadata capacity
MC： Current metadata space size
CCSMN： Minimum compressed class space size
CCSMX： Maximum compressed class space size
CCSC： Current compressed class space size
YGC： The younger generation gc frequency
FGC： Old age GC frequency

New generation garbage collection statistics
jstat -gcnew 109691

S0C： The size of the first surviving area
S1C： The size of the second surviving area
S0U： The size of the first surviving area
S1U： The size of the second surviving area
TT: The number of times an object has survived in a new generation
MTT: The maximum number of times an object has survived in a new generation
DSS: The expected size of the surviving area
EC： The size of Eden Park
EU： The use size of Eden Park
YGC： Garbage collection times of young generation
YGCT： Young generation garbage collection consumes time

New generation memory statistics
jstat -gcnewcapacity 109691

NGCMN： The minimum capacity of the new generation
NGCMX： The largest capacity of the new generation
NGC： Current Cenozoic capacity
S0CMX： The greatest survival 1 Area size
S0C： The current survival 1 Area size
S1CMX： The greatest survival 2 Area size
S1C： The current survival 2 Area size
ECMX： The largest Eden Park
EC： The current size of Eden Park
YGC： Garbage collection times of young generation
FGC： Recovery times of the elderly

Old age garbage collection statistics
jstat -gcold 109691

MC： Method area size
MU： Method area usage size
CCSC: Compress class space size
CCSU: Compressed class space usage size
OC： Old age size
OU： Old age use size
YGC： Garbage collection times of young generation
FGC： Recycling times in the old days
FGCT： Garbage collection used time in the old days
GCT： The total time spent in garbage collection

Memory statistics in the old days
jstat -gcoldcapacity 109691

OGCMN： The smallest capacity in the old days
OGCMX： The biggest capacity in the old days
OGC： The current size of the elderly
OC： Old age size
YGC： Garbage collection times of young generation
FGC： Recycling times in the old days
FGCT： Garbage collection used time in the old days
GCT： The total time spent in garbage collection

In addition to the above use, it can also be executed at intervals jstat -gc 109691 1000 10 1000 Represents milliseconds Namely To perform a second 1 Time , A total of execution 10 Time
You can also export to a file for subsequent viewing jstat -gc 109691 1000 10 >> aa.txt

JVM Operation forecast

use jstat gc -pid Command can calculate the following key data , With this data, you can optimize , Set up some initial... For your system JVM Parameters , Such as heap memory size , Young generation size ,Eden and Survivor The proportion of , The size of the old age , Threshold for large objects , The threshold for older objects to enter the old age, etc .
The rate at which young people grow
Can execute orders jstat -gc pid 1000 10 ( every other 1 Seconds to perform 1 subcommand , In total 10 Time ), Through observation EU(eden The use of the zone ) To estimate... Per second eden About how many new objects , If the system load is not high , You can put the frequency 1 Second to second 1 minute , even to the extent that 10 Minutes to see the whole situation . Be careful , General systems may have peak and daily periods , So we need to estimate the growth rate of objects in different situations at different times .
Young GC The trigger frequency and time consumption of each time
If we know the growth rate of young people, we can infer that eden The size of the area is calculated as Young GC How often does it trigger ,Young GC The average time spent by YGCT/YGC The formula works out , Based on the results, we can probably know how long the system will last because Young GC How long is it going to be stuck .
Every time Young GC After that, how many people survive and enter the elderly generation
This is because I knew about it before Young GC The frequency of , Suppose every 5 Minutes at a time , Then you can execute the command jstat -gc pid 300000 10 , Observe the results every time eden,survivor And changes in the use of older people , In every time gc after eden The use of the zone is generally greatly reduced ,survivor And the elderly are likely to grow , The object of these increases is every time Young GC After the survival of the object , And you can see that every time Young GC How many objects are there in the old age , So we can calculate the growth rate of objects in the old age .
Full GC The trigger frequency and time consumption of each time
Knowing the growth rate of objects in the old age, we can calculate Full GC The trigger frequency of ,Full GC You can use the formula for each time FGCT/FGC calculated .
Optimization idea
In fact, simply put, try to make every time Young GC After the survival of less than Survivor Regional 50%, It's all in the younger generation . Try not to let the object enter the old age . Try to reduce Full GC The frequency of , Avoid frequent Full GC Yes JVM The impact of performance .

Alibaba Arthas Introduction

brief introduction

Arthas yes Alibaba stay 2018 year 9 Open source in Java Diagnostic tools . Support JDK6+, Use command line interaction mode , It is convenient to locate and diagnose online program running problems .Arthas Official documents are very detailed , See ：https://alibaba.github.io/arthas
** Arthas Use scenarios **
Thanks to the Arthas Powerful and rich features , Give Way Arthas What can be done is beyond imagination . Here are just a few common uses , More usage scenarios can be familiar with Arthas Then explore for yourself .

1. Is there a global perspective to see the health of the system ？
2. Why? CPU It's going up again , Where did it take up CPU ？
3. Does the multithread running have a deadlock ？ Is there a jam ？
4. The program takes a long time to run , Where does it take longer ？ How to monitor ？
5. Where does this class come from jar Package loaded ？ Why do you report all kinds of related Exception？
6. Why didn't I change the code to ？ I didn't commit？ The branch is wrong ？
7. I can't get online when I have a problem debug, Can't we just add logs and redistribute them ？
8. Is there any way to monitor JVM The real-time running state of ？

Arthas Start installation

# github download arthas
wget https://alibaba.github.io/arthas/arthas-boot.jar
#  perhaps  Gitee  download 
wget https://arthas.gitee.io/arthas-boot.jar
#  If the download fails, you can go to Arthas  Official download full package   decompression   Then it's running  arthas-boot.jar

Test case code

package com.jvm;
public class Arthas {
    
    private static HashSet hashSet = new HashSet();
    public static void main(String[] args) {
    
        //  simulation  CPU  Too high 
        cpuHigh();
        //  Simulate thread deadlock 
        deadThread();
        //  Keep going to  hashSet  Collection adds data 
        addHashSetThread();
    }
    /** *  Keep going to  hashSet  Set add data  */
    public static void addHashSetThread() {
    
        //  Initializing constants 
        new Thread(() -> {
    
            int count = 0;
            while (true) {
    
                try {
    
                    hashSet.add("count" + count);
                    Thread.sleep(1000);
                    count++;
                } catch (InterruptedException e) {
    
                    e.printStackTrace();
                }
            }
        }).start();
    }
    public static void cpuHigh() {
    
        new Thread(() -> {
    
            while (true) {
    

            }
        }).start();
    }
    /** *  Deadlock  */
    private static void deadThread() {
    
        /**  Create resources  */
        Object resourceA = new Object();
        Object resourceB = new Object();
        //  Create thread 
        Thread threadA = new Thread(() -> {
    
            synchronized (resourceA) {
    
                System.out.println(Thread.currentThread() + " get ResourceA");
                try {
    
                    Thread.sleep(1000);
                } catch (InterruptedException e) {
    
                    e.printStackTrace();
                }
                System.out.println(Thread.currentThread() + "waiting get resourceB");
                synchronized (resourceB) {
    
                    System.out.println(Thread.currentThread() + " get resourceB");
                }
            }
        });

        Thread threadB = new Thread(() -> {
    
            synchronized (resourceB) {
    
                System.out.println(Thread.currentThread() + " get ResourceB");
                try {
    
                    Thread.sleep(1000);
                } catch (InterruptedException e) {
    
                    e.printStackTrace();
                }
                System.out.println(Thread.currentThread() + "waiting get resourceA");
                synchronized (resourceA) {
    
                    System.out.println(Thread.currentThread() + " get resourceA");
                }
            }
        });
        threadA.start();
        threadB.start();
    }
}

Pay attention to the package name when running , Otherwise, the main class will not be found . My side is package com.jvm; stay Linux To create a mkdir -p com/jvm adopt javac Arthas.java compile become .class file And then it's running java com/jvm/Arthas

adopt top Command view

Does cause cpu Gao Sheng .
Next through Arthas solve the problem
function Arthas java -jar arthas-boot.jar

Press 1 Enter the process project

Arthas Common commands

help

Into success adopt help View related commands

dashboard

dashboard： The dashboard

You can see through the dashboard cpu High occupancy and thread blocking

thread

thread: View Thread details
First of all to see cpu High thread

According to the information find com.jvm.Arahas Under this category cpuHigh Method Look at the corresponding code , So as to solve cpu High problem
thread -b You can view thread deadlocks

Then, according to the information, the deadlock problem is solved

jad

Input jad Add the full name of the class Can decompile , This makes it easy for us to check whether the online code is the correct version
jad com.jvm.Arthas

ognl

Arthas 3.0 Use in ognl The expression replaces groovy To implement the evaluation function of the expression , It's solved groovy Potential memory leaks . Flexible use of ognl expression , It can greatly improve the efficiency of troubleshooting .
ognl @[email protected]: View the values in the set

ognl '@[email protected]("testabc123")' : Add values to the collection

More commands can be used help Command view , Or view the document ：https://alibaba.github.io/arthas/commands.html#arthas