- Notifications
You must be signed in to change notification settings - Fork10
xuanbo/easy-java
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
整理java技术要点,让java更简单,更容易上手。
spring-boot
集成默认是2.xMySQL
数据库默认编码为utf8mb4
对于maven项目,推荐使用java8、项目编码为utf8。对于国内的开发者而言,为了下载包更快,推荐使用aliyun仓库。
<?xml version="1.0" encoding="UTF-8"?><projectxmlns="http://maven.apache.org/POM/4.0.0"xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd"> <modelVersion>4.0.0</modelVersion> <groupId>tk.fishfish</groupId> <artifactId>easy-java</artifactId> <version>1.0.0</version> <properties> <maven.compiler.source>1.8</maven.compiler.source> <maven.compiler.target>1.8</maven.compiler.target> <project.build.sourceEncoding>utf-8</project.build.sourceEncoding> </properties> <repositories> <repository> <id>aliyunmaven</id> <name>aliyun maven</name> <url>http://maven.aliyun.com/nexus/content/groups/public/</url> </repository> </repositories> <build> <plugins> <plugin> <groupId>org.apache.maven.plugins</groupId> <artifactId>maven-compiler-plugin</artifactId> <version>3.8.0</version> <configuration> <source>1.8</source> <target>1.8</target> </configuration> </plugin> </plugins> </build></project>
对于多模块依赖项目,请看示例工程
在打包的时候,为了跳过测试,可以使用命令:
> mvn clean package -DskipTests
其中:
-DskipTests
参数用于跳过测试。
IO在工作中会经常遇到,比如文件复制、读写、WEB文件上传/下载等。
建议:工作中尽量使用工具类,能不写new BufferedInputStream(new FileInputStream("file"))
就不写。
对于文件复制,这里不介绍各种FileUtils
、FileCopyUtils
等。
介绍JDK7中优雅的工具类Paths
、Files
使用:
packagetk.fishfish.easyjava.io;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.io.IOException;importjava.nio.file.Files;importjava.nio.file.Path;importjava.nio.file.Paths;/** * 文件复制 * * @author 奔波儿灞 * @since 1.0 */publicclassFileCopyTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(FileCopyTest.class);@Testpublicvoidcopy() {Pathsource =Paths.get("./readme.md");Pathtarget =Paths.get("./readme.md.copy");try {Files.copy(source,target); }catch (IOExceptione) {LOG.error("copy error",e); } }}
Paths
用于构造本地文件路径Files
用于操作文件
所以,下次碰到文件复制相关的需求,试试JDK7标准库呗。
还是优先推荐JDK7标准库Files
:
- 将文件读到字节数组:
Files.readAllBytes
- 将文件一行一行读取:
Files.readAllLines
- 遍历文件夹:
Files.list
- 递归遍历文件夹:
Files.walk
- 将字节数组写入文件:
Files.write
- 复制文件
Files.copy
- ...
如果一些搞不定的操作,再去使用各种工具类吧。
推荐使用apache commons相关包、spring工具类、guava等广泛使用的库。
这里主要介绍spring mvc
中如何使用。
在spring mvc
中通过MultipartFile
去接收单个文件:
packagetk.fishfish.easyjava.controller;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.http.ResponseEntity;importorg.springframework.web.bind.annotation.GetMapping;importorg.springframework.web.bind.annotation.PostMapping;importorg.springframework.web.bind.annotation.RequestPart;importorg.springframework.web.bind.annotation.RestController;importorg.springframework.web.multipart.MultipartFile;importjava.io.IOException;importjava.io.InputStream;importjava.nio.file.Files;importjava.nio.file.Path;importjava.nio.file.Paths;/** * demo Controller * * @author 奔波儿灞 * @since 1.0 */@RestControllerpublicclassDemoController {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(DemoController.class);@PostMapping("/upload")publicResponseEntity<String>upload(@RequestPart("file")MultipartFilefile) {// 文件信息LOG.info("file name: {}",file.getOriginalFilename());LOG.info("file size: {}",file.getSize());// 下面是保存在本地InputStreamin =null;try {Pathtarget =Paths.get("./" +file.getOriginalFilename());in =file.getInputStream();Files.copy(in,target); }catch (IOExceptione) {LOG.error("save file error",e);returnResponseEntity.ok("false"); }finally {if (in !=null) {try {in.close(); }catch (IOExceptione) {LOG.warn("close inputStream error",e); } } }returnResponseEntity.ok("ok"); }}
主要是通过MultipartFile
拿到文件的输入流,之后再balabala。
如果是多文件上传呢?其实将MultipartFile
替换称MultipartFile[]
即可。
主要是将文件字节写入HTTP的响应流,然后设置响应头信息。但是这里不介绍常规的将文件字节写入HttpServletResponse
的方式,而是通过ResponseEntity<byte[]>
:
packagetk.fishfish.easyjava.controller;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.http.HttpHeaders;importorg.springframework.http.HttpStatus;importorg.springframework.http.ResponseEntity;importorg.springframework.web.bind.annotation.GetMapping;importorg.springframework.web.bind.annotation.PostMapping;importorg.springframework.web.bind.annotation.RequestPart;importorg.springframework.web.bind.annotation.RestController;importorg.springframework.web.multipart.MultipartFile;importjava.io.IOException;importjava.io.InputStream;importjava.nio.file.Files;importjava.nio.file.Path;importjava.nio.file.Paths;/** * demo Controller * * @author 奔波儿灞 * @since 1.0 */@RestControllerpublicclassDemoController {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(DemoController.class);@GetMapping("/download")publicResponseEntity<?>download() {Pathsource =Paths.get("./readme.md");try {Stringname =source.toFile().getName();byte[]bytes =Files.readAllBytes(source);HttpHeadersheaders =newHttpHeaders();headers.add("Content-Disposition","attachment;filename=" +name);returnnewResponseEntity<>(bytes,headers,HttpStatus.OK); }catch (IOExceptione) {LOG.error("read file error",e);returnResponseEntity.ok("false"); } }}
原理是一样的,也是设置响应头Content-Disposition
告知浏览器下载。返回ResponseEntity<byte[]>
时,底层还是将文件字节写入HttpServletResponse
的。
这种方式呢,写起来没有直接面向原生的servlet
,完全使用spring mvc
的东西。
建议:在下载较大的文件时,还是推荐使用常规的HttpServletResponse
方式操作。
经常会遇到读取classpath下的某个文件,这时推荐使用spring
工具类ClassPathResource
:
packagetk.fishfish.easyjava.io;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.core.io.ClassPathResource;importjava.io.IOException;/** * classpath * * @author 奔波儿灞 * @since 1.0 */publicclassClassPathTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(ClassPathTest.class);@Testpublicvoidrun()throwsIOException {// maven项目src/main/java、src/test/java、src/main/resourcesClassPathResourceclassPathResource =newClassPathResource("tk/fishfish/easyjava/io/ClassPathTest.class");Stringpath =classPathResource.getFile().getPath();LOG.info("path: {}",path); }}
对于maven项目,如果是运行test cases
,那么如下路径为classpath
:
src/test/java
src/test/resources
如果是运行程序:
src/main/java
src/main/resources
我们经常会碰到判断文件类型的情况,比如,验证用户上传的文件是否为XML
或者其他格式。然后,比较挫的实现方式是判断文件后缀。
哈哈,但是,如果用户故意修改后缀名。那么,是否有办法识别呢?
当时是有的!一般的,几乎所有的可执行文件格式最开始的几个字节都是魔数。为什么呢?你要处理的一段数据,最初进行一个粗略判断。主要是怕处理了半天,原来输入是段垃圾,浪费自己时间,甚至把自己程序搞奔溃。
你可以认为,魔数(magic number)读到是对的,你有就有99%的信心,这个数据应该不是垃圾,值得分析。
比如,JPEG
的文件头为FFD8FF
、PNG
文件头为89504E47
。
这里定义了一个枚举,列举了常见的文件类型,以及魔数。
packagetk.fishfish.easyjava.util;/** * 根据文件魔数判断文件实际类型,防止修改后缀误判 * * @author 奔波儿灞 * @since 1.0 */publicenumFileType {/** * jpg、jpeg */JPEG("JPEG","FFD8FF"),/** * png */PNG("PNG","89504E47"),/** * xml */XML("XML","3C3F786D6C"),/** * html */HTML("HTML","68746D6C3E"),/** * pdf */PDF("PDF","255044462D312E"),/** * zip、jar */ZIP("ZIP","504B0304"),/** * rar */RAR("RAR","52617221"), ;/** * 文件类型 */privateStringtype;/** * 文件魔数 */privateStringvalue;FileType(Stringtype,Stringvalue) {this.type =type;this.value =value; }publicStringgetType() {returntype; }publicStringgetValue() {returnvalue; }}
当然,列举不全,有兴趣自己网上搜索相关资料。
如下是一个工具类,用来根据文件的头来判断文件类型:
packagetk.fishfish.easyjava.util;importjava.io.BufferedInputStream;importjava.io.IOException;importjava.util.Objects;/** * 文件类型工具 * * @author 奔波儿灞 * @since 1.0 */publicfinalclassFileTypeUtils {privatestaticfinalintLEN =28;privateFileTypeUtils() {thrownewIllegalStateException("Utils"); }publicstaticbooleanisType(BufferedInputStreamis,FileTypefileType)throwsIOException {Objects.requireNonNull(fileType);Stringheader =getFileHeader(is);returnheader.startsWith(fileType.getValue()); }privatestaticStringgetFileHeader(BufferedInputStreamis)throwsIOException {Objects.requireNonNull(is);// 标识is.mark(LEN +1);byte[]b =newbyte[LEN];try {intnum =is.read(b,0,LEN);if (num <LEN) {thrownewRuntimeException("can not read file head"); } }finally {// 重置,否则无法重复读取流is.reset(); }returnbytes2hex(b); }privatestaticStringbytes2hex(byte[]bytes) {StringBuilderbuilder =newStringBuilder();for (byteb :bytes) {intv =b &0xFF;Stringhv =Integer.toHexString(v);if (hv.length() <2) {builder.append(0); }builder.append(hv.toUpperCase()); }returnbuilder.toString(); }}
测试如下:
packagetk.fishfish.easyjava.util;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.core.io.ClassPathResource;importjava.io.BufferedInputStream;importjava.io.IOException;/** * 文件类型工具测试 * * @author 奔波儿灞 * @since 1.0 */publicclassFileTypeUtilsTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(FileTypeUtilsTest.class);@TestpublicvoidisXml()throwsIOException {ClassPathResourceresource =newClassPathResource("logback-spring.xml");BufferedInputStreambis =newBufferedInputStream(resource.getInputStream());// 判断是否为XML格式booleanisXml =FileTypeUtils.isType(bis,FileType.XML);LOG.info("isXml: {}",isXml); }}
根据阿里p3c规范,推荐通过线程池来创建线程。而且是手动管理,因为通过Executors
创建的线程池,存在一些缺陷:
FixedThreadPool
和SingleThreadPool
:允许的请求队列长度为Integer.MAX_VALUE
,可能会堆积大量的请求,从而导致OOM。CachedThreadPool
和ScheduledThreadPool
:允许的创建线程数量为Integer.MAX_VALUE
,可能会创建大量的线程,从而导致OOM。
下面是ThreadPoolExecutor
的构造方法:
publicThreadPoolExecutor(intcorePoolSize,intmaximumPoolSize,longkeepAliveTime,TimeUnitunit,BlockingQueue<Runnable>workQueue,ThreadFactorythreadFactory,RejectedExecutionHandlerhandler) {if (corePoolSize <0 ||maximumPoolSize <=0 ||maximumPoolSize <corePoolSize ||keepAliveTime <0)thrownewIllegalArgumentException();if (workQueue ==null ||threadFactory ==null ||handler ==null)thrownewNullPointerException();this.acc =System.getSecurityManager() ==null ?null :AccessController.getContext();this.corePoolSize =corePoolSize;this.maximumPoolSize =maximumPoolSize;this.workQueue =workQueue;this.keepAliveTime =unit.toNanos(keepAliveTime);this.threadFactory =threadFactory;this.handler =handler;}
下面是参数介绍:
参数名 | 说明 |
---|---|
corePoolSize | 核心线程池大小 |
maximumPoolSize | 最大线程池大小 |
keepAliveTime | 线程最大空闲时间 |
unit | 时间单位 |
workQueue | 任务等待队列 |
threadFactory | 线程创建工厂 |
handler | 拒绝策略 |
线程池执行策略:
publicvoidexecute(Runnablecommand) {if (command ==null)thrownewNullPointerException();/* * Proceed in 3 steps: * * 1. If fewer than corePoolSize threads are running, try to * start a new thread with the given command as its first * task. The call to addWorker atomically checks runState and * workerCount, and so prevents false alarms that would add * threads when it shouldn't, by returning false. * * 2. If a task can be successfully queued, then we still need * to double-check whether we should have added a thread * (because existing ones died since last checking) or that * the pool shut down since entry into this method. So we * recheck state and if necessary roll back the enqueuing if * stopped, or start a new thread if there are none. * * 3. If we cannot queue task, then we try to add a new * thread. If it fails, we know we are shut down or saturated * and so reject the task. */intc =ctl.get();if (workerCountOf(c) <corePoolSize) {if (addWorker(command,true))return;c =ctl.get(); }if (isRunning(c) &&workQueue.offer(command)) {intrecheck =ctl.get();if (!isRunning(recheck) &&remove(command))reject(command);elseif (workerCountOf(recheck) ==0)addWorker(null,false); }elseif (!addWorker(command,false))reject(command);}
总结:
- 创建线程直到
corePoolSize
- 任务放入
workQueue
,直到充满 - 继续创建线程直到
maximumPoolSize
- 使用拒绝策略
handler
如下创建线程池:
publicvoidcreate() {// 线程池参数intcorePoolSize =5;intmaximumPoolSize =10;longkeepAliveTime =5;TimeUnitunit =TimeUnit.MINUTES;intworkQueueSize =1000;BlockingQueue<Runnable>workQueue =newLinkedBlockingDeque<>(workQueueSize);ThreadFactorythreadFactory =newDefaultThreadFactory("threadPool");RejectedExecutionHandlerhandler =newThreadPoolExecutor.CallerRunsPolicy();// 创建ExecutorServicethreadPool =newThreadPoolExecutor(corePoolSize,maximumPoolSize,keepAliveTime,unit,workQueue,threadFactory,handler );}
其中DefaultThreadFactory
为ThreadFactory
的一个自定义实现:
packagetk.fishfish.easyjava.threadpool;importjava.util.concurrent.ThreadFactory;importjava.util.concurrent.atomic.AtomicInteger;/** * 默认的ThreadFactory * * @author 奔波儿灞 * @since 1.0 */publicclassDefaultThreadFactoryimplementsThreadFactory {/** * 线程组 */privatefinalThreadGroupgroup;/** * 线程编号 */privatefinalAtomicIntegerthreadNumber =newAtomicInteger(1);/** * 线程池名称前缀 */privatefinalStringnamePrefix;publicDefaultThreadFactory(StringnamePrefix) {SecurityManagermanager =System.getSecurityManager();this.group = (manager !=null) ?manager.getThreadGroup() :Thread.currentThread().getThreadGroup();this.namePrefix =namePrefix +"-"; }@OverridepublicThreadnewThread(Runnablerunnable) {Stringname =namePrefix +threadNumber.getAndIncrement();Threadthread =newThread(group,runnable,name,0);if (thread.isDaemon()) {thread.setDaemon(false); }if (thread.getPriority() !=Thread.NORM_PRIORITY) {thread.setPriority(Thread.NORM_PRIORITY); }returnthread; }}
源码参考JDK默认实现java.util.concurrent.Executors.DefaultThreadFactory
。
JDK默认实现比较适合cpu密集型任务,对于IO密集型任务,线程池执行策略需要调整:
- 优先充满线程直到
maximumPoolSize
- 任务放入
workQueue
,直到充满 - 使用拒绝策略
handler
可以参考新浪RPC框架motan
的实现,其中需要自定义线程池与工作队列。
线程池
StandardThreadExecutor
packagetk.fishfish.easyjava.threadpool;importjava.util.concurrent.RejectedExecutionException;importjava.util.concurrent.RejectedExecutionHandler;importjava.util.concurrent.ThreadFactory;importjava.util.concurrent.ThreadPoolExecutor;importjava.util.concurrent.TimeUnit;importjava.util.concurrent.atomic.AtomicInteger;/*** 优先启动线程** 代码来自motan rpc** @author 奔波儿灞* @since 1.0*/publicclassStandardThreadExecutorextendsThreadPoolExecutor {publicstaticfinalintDEFAULT_MIN_THREADS =20;publicstaticfinalintDEFAULT_MAX_THREADS =200;/** * 1 minutes */publicstaticfinalintDEFAULT_MAX_IDLE_TIME =60 *1000;/** * 正在处理的任务数 */privateAtomicIntegersubmittedTasksCount;/** * 最大允许同时处理的任务数 */privateintmaxSubmittedTaskCount;publicStandardThreadExecutor(intcoreThreads,intmaxThreads,longkeepAliveTime,TimeUnitunit,intqueueCapacity,ThreadFactorythreadFactory,RejectedExecutionHandlerhandler) {super(coreThreads,maxThreads,keepAliveTime,unit,newExecutorQueue(),threadFactory,handler); ((ExecutorQueue)getQueue()).setStandardThreadExecutor(this);submittedTasksCount =newAtomicInteger(0);// 最大并发任务限制: 队列buffer数 + 最大线程数maxSubmittedTaskCount =queueCapacity +maxThreads; }@Overridepublicvoidexecute(Runnablecommand) {intcount =submittedTasksCount.incrementAndGet();// 超过最大的并发任务限制,进行 reject// 依赖的LinkedTransferQueue没有长度限制,因此这里进行控制if (count >maxSubmittedTaskCount) {submittedTasksCount.decrementAndGet();getRejectedExecutionHandler().rejectedExecution(command,this); }try {super.execute(command); }catch (RejectedExecutionExceptionrx) {// there could have been contention around the queueif (!((ExecutorQueue)getQueue()).force(command)) {submittedTasksCount.decrementAndGet();getRejectedExecutionHandler().rejectedExecution(command,this); } } }publicintgetSubmittedTasksCount() {returnthis.submittedTasksCount.get(); }publicintgetMaxSubmittedTaskCount() {returnmaxSubmittedTaskCount; }@OverrideprotectedvoidafterExecute(Runnabler,Throwablet) {submittedTasksCount.decrementAndGet(); }}
任务队列
ExecutorQueue
packagetk.fishfish.easyjava.threadpool;importjava.util.concurrent.LinkedTransferQueue;importjava.util.concurrent.RejectedExecutionException;/*** LinkedTransferQueue 能保证更高性能,相比与LinkedBlockingQueue有明显提升* <p>* 1) 不过LinkedTransferQueue的缺点是没有队列长度控制,需要在外层协助控制* <p>* 代码来自motan rpc** @author 奔波儿灞* @since 1.0*/publicclassExecutorQueueextendsLinkedTransferQueue<Runnable> {privateStandardThreadExecutorthreadPoolExecutor;publicExecutorQueue() {super(); }publicvoidsetStandardThreadExecutor(StandardThreadExecutorthreadPoolExecutor) {this.threadPoolExecutor =threadPoolExecutor; }/** * 注:代码来源于 tomcat * * @param runnable Runnable * @return 是否添加成功 */publicbooleanforce(Runnablerunnable) {if (threadPoolExecutor.isShutdown()) {thrownewRejectedExecutionException("Executor not running, can't force a command into the queue"); }// forces the item onto the queue, to be used if the task is rejectedreturnsuper.offer(runnable); }/** * 注:tomcat的代码进行一些小变更 * 在提交的任务数超过poolSize, 而poolSize小于最大线程数的时候返回false, 让executor创建线程 * * @param runnable Runnable * @return if the runnable was added to this queue */@Overridepublicbooleanoffer(Runnablerunnable) {intpoolSize =threadPoolExecutor.getPoolSize();// we are maxed out on threads, simply queue the objectif (poolSize ==threadPoolExecutor.getMaximumPoolSize()) {returnsuper.offer(runnable); }// we have idle threads, just add it to the queue// note that we don't use getActiveCount(), see BZ 49730if (threadPoolExecutor.getSubmittedTasksCount() <=poolSize) {returnsuper.offer(runnable); }// if we have less threads than maximum force creation of a new threadif (poolSize <threadPoolExecutor.getMaximumPoolSize()) {returnfalse; }// if we reached here, we need to add it to the queuereturnsuper.offer(runnable); }}
具体代码见:
tk.fishfish.easyjava.threadpool
:该包为线程池相关tk.fishfish.easyjava.threadpool.ThreadPoolTest
:线程池测试类
参考:
以spring boot为例,这里主要介绍:
- task:可以认为是quartz的简化版本
- async:将任务提交到线程池,异步执行
使用注解@Scheduled
即可(@EnableScheduling
开启),十分的方便,用于执行一些简单的、固定的任务。
packagetk.fishfish.easyjava.task;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.scheduling.annotation.Scheduled;importorg.springframework.stereotype.Component;/** * 任务 * * @author 奔波儿灞 * @since 1.0 */@ComponentpublicclassTask {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(Task.class);/** * 启动1分钟后,每隔1分钟执行一次 */@Scheduled(initialDelay =1000 *60,fixedDelay =1000 *60)publicvoiddoSomething() {LOG.info("doSomething..."); }}
但是存在的问题是单线程,不能多任务并行执行,因此我们需要自定义任务线程池。
我们只需要创建TaskScheduler
或ScheduledExecutorService
类型的bean即可,spring会将其作为的任务线程池。
2019-05-12 18:02:20.315 INFO 73998 --- [ main] s.a.ScheduledAnnotationBeanPostProcessor : No TaskScheduler/ScheduledExecutorService bean found for scheduled processing
下面是任务的配置,通过ThreadPoolTaskScheduler
,我们可以设置是否等待任务结束再退出程序。
packagetk.fishfish.easyjava.conf.task;importorg.springframework.boot.context.properties.EnableConfigurationProperties;importorg.springframework.context.annotation.Bean;importorg.springframework.context.annotation.Configuration;importorg.springframework.scheduling.TaskScheduler;importorg.springframework.scheduling.annotation.EnableScheduling;importorg.springframework.scheduling.concurrent.ThreadPoolTaskScheduler;importjava.util.concurrent.ThreadPoolExecutor;/** * task配置 * * @author 奔波儿灞 * @since 1.0 */@Configuration@EnableScheduling@EnableConfigurationProperties(TaskProperties.class)publicclassTaskConfiguration {/** * 自定义任务线程池,默认只有一个线程,多个任务无法并发执行 * 需要为TaskScheduler/ScheduledExecutorService类型 * * @return TaskScheduler */@BeanpublicTaskSchedulertaskScheduler(TaskPropertiesproperties) {ThreadPoolTaskSchedulertaskScheduler =newThreadPoolTaskScheduler();// 设置线程数taskScheduler.setPoolSize(properties.getPoolSize());// 设置默认线程名称taskScheduler.setThreadNamePrefix(properties.getThreadNamePrefix());// 等待所有任务结束后再关闭线程池taskScheduler.setWaitForTasksToCompleteOnShutdown(properties.isWaitForTasksToCompleteOnShutdown());// 设置拒绝策略taskScheduler.setRejectedExecutionHandler(newThreadPoolExecutor.CallerRunsPolicy());returntaskScheduler; }}
通过自定义TaskProperties
,我们可以通过配置文件定义线程池参数,而不需要改代码。
packagetk.fishfish.easyjava.conf.task;importorg.springframework.boot.context.properties.ConfigurationProperties;/** * task properties * * @author 奔波儿灞 * @since 1.0 */@ConfigurationProperties(TaskProperties.PREFIX)publicclassTaskProperties {publicstaticfinalStringPREFIX ="task";/** * 线程数 */privateintpoolSize =1;/** * 是否等待任务结束再退出 */privatebooleanwaitForTasksToCompleteOnShutdown =true;/** * 线程前缀 */privateStringthreadNamePrefix =PREFIX;publicintgetPoolSize() {returnpoolSize; }publicvoidsetPoolSize(intpoolSize) {this.poolSize =poolSize; }publicbooleanisWaitForTasksToCompleteOnShutdown() {returnwaitForTasksToCompleteOnShutdown; }publicvoidsetWaitForTasksToCompleteOnShutdown(booleanwaitForTasksToCompleteOnShutdown) {this.waitForTasksToCompleteOnShutdown =waitForTasksToCompleteOnShutdown; }publicStringgetThreadNamePrefix() {returnthreadNamePrefix; }publicvoidsetThreadNamePrefix(StringthreadNamePrefix) {this.threadNamePrefix =threadNamePrefix; }}
例如,在配置文件中:
# 任务线程池task:pool-size:2wait-for-tasks-to-complete-on-shutdown:truethread-name-prefix:taskPool-
具体代码见:
tk.fishfish.easyjava.conf.task
:该包下为任务配置相关tk.fishfish.easyjava.task.Task
:任务类(demo)tk.fishfish.easyjava.task.TaskTest
:任务测试
我们可以通过在方法上面添加@Async
,将方法异步化(@EnableAsync
注解开启异步)。即方法会提交到异步线程池中执行,比较适合耗时的任务,而前端又需要立即返回。
@ServicepublicclassAsyncServiceImplimplementsAsyncService {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(AsyncServiceImpl.class);@Override@AsyncpublicvoiddoSomething() {// 做些什么LOG.info("doSomething..."); }}
当然,与task类似的是,我们也可以自定义异步执行的线程池。
其实只要配置一个类型为TaskExecutor
,bean的名称为taskExecutor
的Bean即可。
packagetk.fishfish.easyjava.conf.async;importorg.springframework.boot.context.properties.EnableConfigurationProperties;importorg.springframework.context.annotation.Bean;importorg.springframework.context.annotation.Configuration;importorg.springframework.core.task.TaskExecutor;importorg.springframework.scheduling.annotation.EnableAsync;importorg.springframework.scheduling.concurrent.ThreadPoolTaskExecutor;importjava.util.concurrent.ThreadPoolExecutor;/** * async配置 * * @author 奔波儿灞 * @since 1.0 */@Configuration@EnableAsync@EnableConfigurationProperties(AsyncProperties.class)publicclassAsyncConfiguration {/** * 类型为`TaskExecutor`,bean的名称为`taskExecutor`的Bean * * @return TaskExecutor */@BeanpublicTaskExecutortaskExecutor(AsyncPropertiesproperties) {ThreadPoolTaskExecutorexecutor =newThreadPoolTaskExecutor();// 线程池前缀,可以随意指定。executor.setThreadNamePrefix(properties.getThreadNamePrefix());// 设置线程池参数executor.setCorePoolSize(properties.getCorePoolSize());executor.setMaxPoolSize(properties.getMaxPoolSize());executor.setQueueCapacity(properties.getQeueCapacity());executor.setKeepAliveSeconds(properties.getKeepAliveSeconds());executor.setWaitForTasksToCompleteOnShutdown(properties.isWaitForTasksToCompleteOnShutdown());// 设置拒绝策略,由调用者执行executor.setRejectedExecutionHandler(newThreadPoolExecutor.CallerRunsPolicy());returnexecutor; }}
其中,AsyncProperties
封装了线程池相关参数,方便使用配置文件灵活配置:
packagetk.fishfish.easyjava.conf.async;importorg.springframework.boot.context.properties.ConfigurationProperties;/** * async properties * * @author 奔波儿灞 * @since 1.0 */@ConfigurationProperties(AsyncProperties.PREFIX)publicclassAsyncProperties {publicstaticfinalStringPREFIX ="async";/** * 核心线程数 */privateintcorePoolSize =1;/** * 最大线程数 */privateintmaxPoolSize =10;/** * 队列容量 */privateintqueueCapacity =100;/** * 线程空闲时间 */privateintkeepAliveSeconds =60;/** * 是否等待任务结束再退出 */privatebooleanwaitForTasksToCompleteOnShutdown =true;/** * 线程前缀 */privateStringthreadNamePrefix =PREFIX;publicintgetCorePoolSize() {returncorePoolSize; }publicvoidsetCorePoolSize(intcorePoolSize) {this.corePoolSize =corePoolSize; }publicintgetMaxPoolSize() {returnmaxPoolSize; }publicvoidsetMaxPoolSize(intmaxPoolSize) {this.maxPoolSize =maxPoolSize; }publicintgetQueueCapacity() {returnqueueCapacity; }publicvoidsetQueueCapacity(intqueueCapacity) {this.queueCapacity =queueCapacity; }publicintgetKeepAliveSeconds() {returnkeepAliveSeconds; }publicvoidsetKeepAliveSeconds(intkeepAliveSeconds) {this.keepAliveSeconds =keepAliveSeconds; }publicbooleanisWaitForTasksToCompleteOnShutdown() {returnwaitForTasksToCompleteOnShutdown; }publicvoidsetWaitForTasksToCompleteOnShutdown(booleanwaitForTasksToCompleteOnShutdown) {this.waitForTasksToCompleteOnShutdown =waitForTasksToCompleteOnShutdown; }publicStringgetThreadNamePrefix() {returnthreadNamePrefix; }publicvoidsetThreadNamePrefix(StringthreadNamePrefix) {this.threadNamePrefix =threadNamePrefix; }}
例如,在配置文件中:
# async线程池async:core-pool-size:2max-pool-size:2queue-capacity:100keep-alive-seconds:60wait-for-tasks-to-complete-on-shutdown:truethread-name-prefix:asyncPool-
具体代码见:
tk.fishfish.easyjava.conf.async
:该包下为async配置相关tk.fishfish.easyjava.async
:该包下为async测试相关
完整探究过程,看我整理的Spring Boot使用@Async即可。
Future
是Java5增加的类,它用来描述一个异步计算的结果。你可以使用isDone
方法检查计算是否完成,或者使用get
方法阻塞住调用线程,直到计算完成返回结果。你也可以使用cancel
方法停止任务的执行。
下面来一个栗子:
packagetk.fishfish.easyjava.threadpool;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.ExecutionException;importjava.util.concurrent.ExecutorService;importjava.util.concurrent.Executors;importjava.util.concurrent.Future;/** * JDK Future * * @author 奔波儿灞 * @since 1.0 */publicclassFutureTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(FutureTest.class);@Testpublicvoidrun() {ExecutorServicethreadPool =Executors.newFixedThreadPool(1);Future<Integer>future =threadPool.submit(() -> {Thread.sleep(10000);// 结果return100; });// do somethingtry {// main阻塞等待结果Integerresult =future.get();LOG.info("result: {}",result); }catch (InterruptedException |ExecutionExceptione) {LOG.warn("获取结果异常",e); } }}
在这个例子中,我们往线程池中提交了一个任务并立即返回了一个Future
对象,接着可以做一些其他操作,最后利用它的get
方法阻塞等待结果或isDone
方法轮询等待结果。
虽然这些方法提供了异步执行任务的能力,但是对于结果的获取却还是很不方便,只能通过阻塞或者轮询的方式得到任务的结果。
阻塞的方式显然和我们的异步编程的初衷相违背,轮询的方式又会耗费无谓的CPU资源,而且也不能及时的得到计算结果,为什么不能用观察者设计模式当计算结果完成及时通知监听者呢?
很多语言,比如Node.js
,采用Callback
的方式实现异步编程。Java
的一些框架,比如Netty
,自己扩展了Java
的Future
接口,提供了addListener
等多个扩展方法。Google
的guava
也提供了通用的扩展Future
:ListenableFuture
、SettableFuture
以及辅助类Futures
等,方便异步编程。为此,Java
终于在JDK1.8
这个版本中增加了一个能力更强的Future
类:CompletableFuture
。它提供了非常强大的Future
的扩展功能,可以帮助我们简化异步编程的复杂性,提供了函数式编程的能力,可以通过回调的方式处理计算结果。下面来看看这几种方式。
引入Maven依赖:
<dependency> <groupId>io.netty</groupId> <artifactId>netty-all</artifactId> <version>4.1.29.Final</version></dependency>
测试代码:
packagetk.fishfish.easyjava.threadpool;importio.netty.util.concurrent.DefaultEventExecutorGroup;importio.netty.util.concurrent.EventExecutorGroup;importio.netty.util.concurrent.Future;importio.netty.util.concurrent.FutureListener;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.CountDownLatch;/** * Netty-Future * * @author 奔波儿灞 * @since 1.0 */publicclassNettyFutureTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(NettyFutureTest.class);@Testpublicvoidrun() {EventExecutorGroupgroup =newDefaultEventExecutorGroup(4);LOG.info("开始");finalCountDownLatchlatch =newCountDownLatch(1);Future<Integer>f =group.submit(() -> {LOG.info("开始耗时计算");Thread.sleep(10000);LOG.info("结束耗时计算");return100; });f.addListener((FutureListener<Integer>)future -> {LOG.info("计算结果: {}",future.get());latch.countDown(); });LOG.info("结束");try {// 不让守护线程退出latch.await(); }catch (InterruptedExceptione) {LOG.warn("等待异常",e); } }}
运行结果:
14:49:06.702 [main] DEBUG io.netty.util.internal.logging.InternalLoggerFactory - Using SLF4J as the default logging framework14:49:06.732 [main] INFO tk.fishfish.easyjava.threadpool.NettyFutureTest - 开始14:49:06.791 [defaultEventExecutorGroup-2-1] INFO tk.fishfish.easyjava.threadpool.NettyFutureTest - 开始耗时计算14:49:06.792 [main] INFO tk.fishfish.easyjava.threadpool.NettyFutureTest - 结束14:49:16.796 [defaultEventExecutorGroup-2-1] INFO tk.fishfish.easyjava.threadpool.NettyFutureTest - 结束耗时计算14:49:16.799 [defaultEventExecutorGroup-2-1] DEBUG io.netty.util.internal.InternalThreadLocalMap - -Dio.netty.threadLocalMap.stringBuilder.initialSize: 102414:49:16.800 [defaultEventExecutorGroup-2-1] DEBUG io.netty.util.internal.InternalThreadLocalMap - -Dio.netty.threadLocalMap.stringBuilder.maxSize: 409614:49:16.801 [defaultEventExecutorGroup-2-1] INFO tk.fishfish.easyjava.threadpool.NettyFutureTest - 计算结果: 100
可以发现,守护线程已经运行完了,当线程池任务执行完成后,回调获取计算是在defaultEventExecutorGroup-2-1
线程中执行的,避免了守护线程无谓的等待。
首先引入guava的Maven依赖:
<dependency> <groupId>com.google.guava</groupId> <artifactId>guava</artifactId> <version>27.1-jre</version></dependency>
测试代码:
packagetk.fishfish.easyjava.threadpool;importcom.google.common.util.concurrent.FutureCallback;importcom.google.common.util.concurrent.Futures;importcom.google.common.util.concurrent.ListenableFuture;importcom.google.common.util.concurrent.ListeningExecutorService;importcom.google.common.util.concurrent.MoreExecutors;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.CountDownLatch;importjava.util.concurrent.ExecutorService;importjava.util.concurrent.Executors;/** * Guava Future * @author 奔波儿灞 * @since 1.0 */publicclassGuavaFutureTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(GuavaFutureTest.class);@Testpublicvoidrun() {LOG.info("开始");ExecutorServiceexecutorService =Executors.newFixedThreadPool(10);ListeningExecutorServiceservice =MoreExecutors.listeningDecorator(executorService);finalCountDownLatchlatch =newCountDownLatch(1);ListenableFuture<Integer>future =service.submit(() -> {LOG.info("开始耗时计算");Thread.sleep(10000);LOG.info("结束耗时计算");return100; });Futures.addCallback(future,newFutureCallback<Integer>() {@OverridepublicvoidonSuccess(Integerresult) {LOG.info("成功,计算结果: {}",result);latch.countDown(); }@OverridepublicvoidonFailure(Throwablethrowable) {LOG.warn("失败",throwable);latch.countDown(); } },executorService);LOG.info("结束");try {// 不让守护线程退出latch.await(); }catch (InterruptedExceptione) {LOG.warn("等待异常",e); } }}
运行结果:
15:16:48.703 [main] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 开始15:16:48.767 [pool-1-thread-1] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 开始耗时计算15:16:48.773 [main] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 结束15:16:58.771 [pool-1-thread-1] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 结束耗时计算15:16:58.775 [pool-1-thread-2] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 成功,计算结果: 100
跟Netty
类似,回调计算获取结果是在pool-1-thread-2
线程,避免了守护线程无谓的等待。
前面提到了Netty
和Guava
的扩展都提供了addListener
这样的接口,用于处理Callback
调用,但其实JDK8
已经提供了一种更为高级的回调方式:CompletableFuture
。
测试代码:
packagetk.fishfish.easyjava.threadpool;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.CompletableFuture;importjava.util.concurrent.CountDownLatch;/** * CompletableFuture * * @author 奔波儿灞 * @since 1.0 */publicclassCompletableFutureTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(CompletableFutureTest.class);@Testpublicvoidrun() {LOG.info("开始");finalCountDownLatchlatch =newCountDownLatch(1);CompletableFuture<Integer>completableFuture =CompletableFuture.supplyAsync(() -> {LOG.info("开始耗时计算");try {Thread.sleep(10000); }catch (InterruptedExceptione) {LOG.warn("执行异常",e); }LOG.info("结束耗时计算");return100; });completableFuture.whenComplete((result,e) -> {LOG.info("回调结果: {}",result);latch.countDown(); });LOG.info("结束");try {// 不让守护线程退出latch.await(); }catch (InterruptedExceptione) {LOG.warn("等待异常",e); } }}
运行结果:
15:25:01.980 [main] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 开始15:25:02.037 [ForkJoinPool.commonPool-worker-9] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 开始耗时计算15:25:02.038 [main] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 结束15:25:12.041 [ForkJoinPool.commonPool-worker-9] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 结束耗时计算15:25:12.041 [ForkJoinPool.commonPool-worker-9] INFO tk.fishfish.easyjava.threadpool.GuavaFutureTest - 回调结果: 100
跟Netty
和Guava
类似,回调计算获取结果是在ForkJoinPool.commonPool-worker-9
线程,避免了守护线程无谓的等待。
关于spring-boot
,网上有太多的教程,这里只介绍一些基本的操作。
对于web项目,依赖spring-boot-starter-web
:
<dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-web</artifactId> <version>${spring.boot.version}</version></dependency>
启动入口:
packagetk.fishfish.easyjava;importorg.springframework.boot.SpringApplication;importorg.springframework.boot.autoconfigure.SpringBootApplication;/** * 入口 * * @author 奔波儿灞 * @since 1.0 */@SpringBootApplicationpublicclassApplication {publicstaticvoidmain(String[]args) {SpringApplication.run(Application.class,args); }}
写个controller:
packagetk.fishfish.easyjava.controller;importorg.springframework.http.ResponseEntity;importorg.springframework.web.bind.annotation.GetMapping;importorg.springframework.web.bind.annotation.RestController;/** * demo Controller * * @author 奔波儿灞 * @since 1.0 */@RestControllerpublicclassDemoController {@GetMappingpublicResponseEntity<String>echo() {returnResponseEntity.ok("easy java"); }}
这样一个REST服务接口就完成了,接下来你需要自行阅读其他文章深造,这里推荐:
- 纯洁的微笑
- spring-boot-guider,这个是我总结的spring-boot(基于1.5.10.RELEAS)入门,适合学习mybatis、reids、mongodb集成
在spring-boot
中有一句话,没有什么是一个注解搞不定的。如果有,那么再加一个。
本次内容参考自关于注解我们需要知道的,建议阅读一番。
从JDK5
开始增加了对注解的支持,注解可以在编译,类加载和运行时被读取,并执行相应一些定义好的处理。通过注解可以在不改变原有代码和逻辑的情况下进行一些其他的补充操作。
在java中系统为我们预置了一部分注解,我们可以通过这些注解来定义其他注解的作用和有效范围等特性。
@Target用于说明Annotation
所修饰的对象范围,所能修饰的范围都被定义在枚举类ElementType
中。
publicenumElementType {// 表示可以用于类,接口,注解或者枚举定义中TYPE,// 字段FIELD,// 方法(不包括构造方法)METHOD,// 方法的参数PARAMETER,// 构造方法上CONSTRUCTOR,// 局部变量LOCAL_VARIABLE,// 只能用在注解上ANNOTATION_TYPE,// 作用包上 package-info.javaPACKAGE,// 表示注解能写在类型变量(泛型参数)的声明语句中如 List<Integer> list = new @Save ArrayList<>();TYPE_PARAMETER,// 表示注解能写在使用类型的任何语句中(声明语句、泛型和强制转换语句中的类型TYPE_USE}
Retention
定义了该Annotation
被保留的时间长短:表示需要在什么级别保存注解信息,用于描述注解的生命周期(即被描述的注解在什么范围内有效),取值被定义在枚举类RetentionPolicy
中:
publicenumRetentionPolicy {// 表示在源代码时有效,编译后的文件没有该注解,一般该类注解仅用于标识如@SuppressWarningsSOURCE,// 默认行为 自定义注解如果没有显示的声明则默认为该行为 在编译时不会被抛弃,但是会被虚拟机抛弃CLASS,// 保留到运行时,可以通过反射来获取 一般该类注解会影响系统的运行RUNTIME}
@Documented@Retention(RetentionPolicy.RUNTIME)@Target(ElementType.ANNOTATION_TYPE)public @interfaceDocumented {}
从注解定义可以看到该注解用在注解定义上。
@Documented
用于描述其它类型的Annotation
应该被作为被标注的程序成员的公共API,因此可以被如javadoc
之类的工具文档化。但是实际使用并不多,有其他更好的替代。
@Inherited
是一个标记注解,@Inherited
表示被其标注的类型是被继承的。如果一个使用了@Inherited
修饰的Annotation
类型被用于一个class
,则这个Annotation
将被用于该class
的子类。
问题:那么注解作用于方法上,子类重写该方法,能获取到吗?
上面介绍的几种元注解是在我们进行自定义注解的时候会用到的,而下面我们介绍几种平时业务开发会经常使用的注解。
@Deprecated
用来描述在当前系统中已经被废弃不推荐使用的类或方法等。
@Target(value={CONSTRUCTOR,FIELD,LOCAL_VARIABLE,METHOD,PACKAGE,PARAMETER,TYPE})
如果我们使用了被@Deprecated
标注的类或方法等,在进行编译的时候会显示相应的提示信息。
@Override
是我们使用很频繁的一个注解,由于重写的操作仅存在于方法中,所以@Override
也只能对方法进行标注。
@Override
功能主要是用来校验当前被标注的方法是否为重写方法,平时我们在继承抽象类或实现接口时都应使用该注解来标注被重写的方法。
@SuppressWarnings
用于可选择的抑制编译器在编译时产生警告信息。
@Target({TYPE,FIELD,METHOD,PARAMETER,CONSTRUCTOR,LOCAL_VARIABLE})
@SuppressWarnings
可选择的值有很多:
deprecation
:不产生使用过期方法(...)的警告,@SuppressWarnings("deprecation")unchecked
:执行了未检查的转换的警告finally
:finally语句无法正常完成时的警告- ...
all
:任意类型的警告
自定义一个注解及其简单,使用@interface
关键字即可完成。同时我们需要确定我们定义的注解使用范围和其具体用途,根据此来确定使用元注解的哪些参数来修饰我们定义的注解。
这里我们定义一个@Log
注解用于记录方法调用日志。
packagetk.fishfish.easyjava.annotation;importjava.lang.annotation.Documented;importjava.lang.annotation.ElementType;importjava.lang.annotation.Retention;importjava.lang.annotation.RetentionPolicy;importjava.lang.annotation.Target;/** * 自定义日志注解,用于记录方法调用时日志 * * @author 奔波儿灞 * @since 1.0 */@Documented@Retention(RetentionPolicy.RUNTIME)@Target({ElementType.METHOD})public @interfaceLog {/** * 模块名称 * * @return 模块名称 */Stringmodule();/** * 功能名称 * * @return 功能名称 */Stringfunction();/** * 描述 * * @return 描述 */Stringdescription();}
使用注解:
packagetk.fishfish.easyjava.annotation;/** * 没有卵用的service实现 * * @author 奔波儿灞 * @since 1.0 */publicclassSomeServiceImplimplementsSomeService {@Override// 使用我们定义的注解@Log(module ="some",function ="find",description ="根据id查询")publicSomefindById(Longid) {// 模拟一些操作if (id %2 ==0) {returnnull; }returnnewSome(id,"随机"); }}
测试一下:
packagetk.fishfish.easyjava.annotation;importorg.junit.Before;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;/** * 测试 * * @author 奔波儿灞 * @since 1.0 */publicclassSomeServiceTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(SomeServiceTest.class);privateSomeServicesomeService;@Beforepublicvoidsetup() {someService =newSomeServiceImpl(); }@TestpublicvoidfindById() {Somesome =someService.findById(1L);LOG.info("some: {}",some); }}
运行该测试用例会发现,并没有什么卵用。因为,还没有自定义解析规则。
16:28:43.366 [main] INFO tk.fishfish.easyjava.annotation.SomeServiceTest - some: Some{id=1, name='随机'}
测试代码见:
tk.fishfish.easyjava.annotation.Log
:自定义注解tk.fishfish.easyjava.annotation.Some
:javabeantk.fishfish.easyjava.annotation.SomeService
:服务接口tk.fishfish.easyjava.annotation.SomeServiceImpl
:服务实现tk.fishfish.easyjava.annotation.SomeServiceTest
:测试用例
JDK提供InvocationHandler
和Proxy
,用于动态代理。基于此,我们拦截到方法,判断注解,然后进行逻辑处理。
packagetk.fishfish.easyjava.annotation;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.lang.reflect.InvocationHandler;importjava.lang.reflect.Method;importjava.lang.reflect.Proxy;/** * Log代理 * * @author 奔波儿灞 * @since 1.0 */publicclassLogProxyimplementsInvocationHandler {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(LogProxy.class);/** * 被代理的类 */privateObjecttarget;@SuppressWarnings("unchecked")public <T>Tbind(Ttarget) {this.target =target;return (T)Proxy.newProxyInstance(target.getClass().getClassLoader(),target.getClass().getInterfaces(),this); }@OverridepublicObjectinvoke(Objectproxy,Methodmethod,Object[]args)throwsThrowable {// 这一步获取实际类的方法,因为method是接口的方法时,是获取不到实现类上的注解信息的MethodrealMethod =getRealMethod(method);// 查找方法上是否存在该注解Loglog =realMethod.getDeclaredAnnotation(Log.class);if (log ==null) {LOG.debug("方法: {} 无@Log注解",method); }else {Stringmodule =log.module();Stringfunction =log.function();Stringdescription =log.description();// 这里我们可以保存到数据库,或者怎么样LOG.info("module: {}, function: {}, description: {}",module,function,description); }// 反射运行方法,返回运行结果returnmethod.invoke(target,args); }privateMethodgetRealMethod(Methodmethod)throwsNoSuchMethodException {returntarget.getClass().getMethod(method.getName(),method.getParameterTypes()); }}
其中getRealMethod
方法用于获取实际类的方法,因为invoke
传入的method
对象是接口的方法,接口上是没有@Log
注解的。
测试:
packagetk.fishfish.easyjava.annotation;importorg.junit.Before;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;/** * 测试 * * @author 奔波儿灞 * @since 1.0 */publicclassSomeServiceTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(SomeServiceTest.class);privateSomeServicesomeService;privateSomeServiceproxySomeService;@Beforepublicvoidsetup() {someService =newSomeServiceImpl();// 代理proxySomeService =newLogProxy().bind(someService); }@TestpublicvoidfindById() {Somesome =someService.findById(1L);LOG.info("some: {}",some); }@TestpublicvoidfindByIdProxy() {Somesome =proxySomeService.findById(1L);LOG.info("some: {}",some); }}
运行findByIdProxy
测试方法即可看到打印注解上获取的信息:
17:38:23.359 [main] INFO tk.fishfish.easyjava.annotation.LogProxy - module: some, function: find, description: 根据id查询17:38:23.363 [main] INFO tk.fishfish.easyjava.annotation.SomeServiceTest - some: Some{id=1, name='随机'}
通过这个例子,大家应该可以了解到JDK动态代理是需要定义接口的。
其实还有一种是基于子类字节码实现的方案——cglib
,使用起来比较复杂,我们一般使用spring aop
简化动态代理。
建议阅读下官方文档,网上各种博客参差不齐,容易进坑。
经过阅读一番,大概知道一些相关的注解:
@Aspect
:申明切面类@Pointcut
:切面表达式(execution、@annotation等)@Before
:前置通知@After
:后置通知- ...
在spring-boot
中,只需要依赖:
<dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-aop</artifactId> <version>${spring.boot.version}</version></dependency>
还是上面的@Log
注解,类似JDK实现的LogProxy
,我们通过AOP的编程方式去解析:
packagetk.fishfish.easyjava.annotation;importorg.aspectj.lang.JoinPoint;importorg.aspectj.lang.annotation.Aspect;importorg.aspectj.lang.annotation.Before;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.stereotype.Component;/** * `@Log`切面 * * @author 奔波儿灞 * @since 1.0 */@Aspect@ComponentpublicclassLogAspect {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(LogAspect.class);@Before(value ="@annotation(log)",argNames ="log")publicvoidbefore(JoinPointjoinPoint,Loglog) {Stringmodule =log.module();Stringfunction =log.function();Stringdescription =log.description();// 这里我们可以保存到数据库,或者怎么样LOG.info("module: {}, function: {}, description: {}",module,function,description); }}
通过@Before
前置通知,在方法调用前记录日志。
测试代码:
packagetk.fishfish.easyjava.annotation;importorg.junit.Test;importorg.junit.runner.RunWith;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.beans.factory.annotation.Autowired;importorg.springframework.boot.test.context.SpringBootTest;importorg.springframework.test.context.junit4.SpringRunner;/** * AOP测试 * * @author 奔波儿灞 * @since 1.0 */@SpringBootTest@RunWith(SpringRunner.class)publicclassSomeServiceAopTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(SomeServiceAopTest.class);@AutowiredprivateSomeServicesomeService;@TestpublicvoidfindById() {Somesome =someService.findById(1L);LOG.info("some: {}",some); }}
注意:上面的SomeServiceImpl
需要增加@Service
注解。
完整的测试代码见:
tk.fishfish.easyjava.annotation
:该包为AOP相关的测试代码
写在最后,对于AOP的其他骚操作,这里不做过多介绍。建议阅读官方文档,参考官方例子,然后本地亲自试一试。
我们应该依赖日志接口,而不是具体的日志实现。这样方便后期更换其他实现,而不需要改代码,虽然我们一般不会更改其他实现。哈哈!
我们应该使用slf4j-api
这个日志接口,再根据实际情况选择log4j
或logback
作为日志的实现。
这里主要以logback
为例:
<!-- https://mvnrepository.com/artifact/org.slf4j/slf4j-api--><dependency> <groupId>org.slf4j</groupId> <artifactId>slf4j-api</artifactId> <version>1.7.25</version></dependency><!-- https://mvnrepository.com/artifact/ch.qos.logback/logback-classic--><dependency> <groupId>ch.qos.logback</groupId> <artifactId>logback-classic</artifactId> <version>1.2.3</version></dependency>
由于logback
已经依赖了slf4j-api
,所以不需要再添加slf4j-api
的依赖。这里为了演示,还是添加了。
注意2点即可:
- 导入
slf4j-api
的包 - 占位符
{}
如下是一个示例:
packagetk.fishfish.easyjava.log;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;/** * 测试日志 * * @author 奔波儿灞 * @since 1.0 */publicclassLogTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(LogTest.class);@Testpublicvoiddebug() {LOG.debug("debug示例,参数:{}","值"); }}
这里仍以logback
为例。
默认情况下,上面的LogTest
测试打印如下信息:
10:54:17.709 [main] DEBUG tk.fishfish.easyjava.log.LogTest - debug示例,参数:值
我们可能想去自定义日志格式,具体的配置见logback
的官网或者一些博客即可。
下面主要介绍logback
异步打印日志,并保存15天,日志文件最大50MB,是我在生产上用的常用的配置:
<?xml version="1.0" encoding="UTF-8"?><configurationdebug="false"><!-- logback配置--><!-- 日志保存路径--> <propertyname="logger.home"value="logs"/><!-- 日志文件名称--> <propertyname="logger.app"value="easy-java"/><!-- 日志级别--> <propertyname="logger.level"value="INFO"/><!-- 日志appender:STDOUT、FILE、ASYNC_FILE--> <propertyname="logger.appender"value="ASYNC_FILE"/><!-- 标准输出--> <appendername="STDOUT"class="ch.qos.logback.core.ConsoleAppender"> <encoderclass="ch.qos.logback.classic.encoder.PatternLayoutEncoder"><!--格式化输出:%-5level:级别从左显示5个字符宽度,%d表示日期,%thread表示线程名,%-50logger{50}:输入方法,%msg:日志消息,%n是换行符--> <pattern>%highlight(%-5level) %d{yyyy-MM-dd HH:mm:ss.SSS} [%thread] %cyan(%-50logger{50}) - %highlight(%msg%n)</pattern> </encoder> </appender><!-- 按照每天生成日志文件--> <appendername="FILE"class="ch.qos.logback.core.rolling.RollingFileAppender"> <rollingPolicyclass="ch.qos.logback.core.rolling.SizeAndTimeBasedRollingPolicy"><!--日志文件输出的文件名--> <FileNamePattern>${logger.home}/${logger.app}.%d{yyyy-MM-dd}.%i.log</FileNamePattern><!--日志文件保留天数--> <MaxHistory>15</MaxHistory><!--日志文件最大的大小--> <MaxFileSize>50MB</MaxFileSize> </rollingPolicy> <encoderclass="ch.qos.logback.classic.encoder.PatternLayoutEncoder"><!--格式化输出:%-5level:级别从左显示5个字符宽度,%d表示日期,%thread表示线程名,%-50logger{50}:输入方法,%msg:日志消息,%n是换行符--> <pattern>%highlight(%-5level) %d{yyyy-MM-dd HH:mm:ss.SSS} [%thread] %cyan(%-50logger{50}) - %highlight(%msg%n)</pattern> </encoder> </appender><!-- 异步输出日志--> <appendername="ASYNC_FILE"class="ch.qos.logback.classic.AsyncAppender"><!-- 不丢失日志.默认的,如果队列的80%已满,则会丢弃TRACT、DEBUG、INFO级别的日志--> <discardingThreshold>0</discardingThreshold><!-- 更改默认的队列的深度,该值会影响性能.默认值为256--> <queueSize>1024</queueSize><!-- 添加附加的appender,最多只能添加一个--> <appender-refref="FILE"/> </appender><!-- 日志输出级别--> <rootlevel="${logger.level}"> <appender-refref="${logger.appender}"/> </root></configuration>
具体配置见:
/src/main/resources/logback-spring.xml
:日志配置文件
在spring-boot中则配置如下即可:
# 日志配置logging:level:# 全局日志级别root:info# 日志配置文件config:classpath:logback-spring.xml
主要是通过logging.config
指定配置文件的路径。
写代码的时候,还是不要太自信,写点单元测试测试下。
平时用的最为广泛的就是junit
了。
<dependency> <groupId>junit</groupId> <artifactId>junit</artifactId> <version>4.12</version> <scope>test</scope></dependency>
下面是简单的使用:
packagetk.fishfish.easyjava.threadpool;importorg.junit.After;importorg.junit.Before;importorg.junit.Test;importjava.util.concurrent.BlockingQueue;importjava.util.concurrent.ExecutorService;importjava.util.concurrent.LinkedBlockingDeque;importjava.util.concurrent.RejectedExecutionHandler;importjava.util.concurrent.ThreadFactory;importjava.util.concurrent.ThreadPoolExecutor;importjava.util.concurrent.TimeUnit;/** * 线程池测试 * * @author 奔波儿灞 * @since 1.0 */publicclassThreadPoolTest {privateExecutorServicethreadPool;@Beforepublicvoidsetup() {intcorePoolSize =5;intmaxPoolSize =10;longkeepAliveTime =5;TimeUnitunit =TimeUnit.MINUTES;intworkQueueSize =1000;BlockingQueue<Runnable>workQueue =newLinkedBlockingDeque<>(workQueueSize);ThreadFactorythreadFactory =newDefaultThreadFactory("threadPool");RejectedExecutionHandlerhandler =newThreadPoolExecutor.CallerRunsPolicy();threadPool =newThreadPoolExecutor(corePoolSize,maxPoolSize,keepAliveTime,unit,workQueue,threadFactory,handler ); }@Testpublicvoidrun()throwsInterruptedException {threadPool.execute(() ->System.out.println("run!!!"));// 为了等待线程池执行完Thread.sleep(3 *1000); }@Afterpublicvoidcleanup() {threadPool.shutdown(); }}
其中:
@Test
注解是测试的具体方法@Before
注解会在初始调用,用于初始化@After
注解会在结束调用,用于清理资源
具体代码见:
tk.fishfish.easyjava.threadpool.ThreadPoolTest
:线程池测试
在spring-boot中使用测试比较简单,需要添加spring-boot-starter-test
依赖:
<dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-test</artifactId> <version>${spring.boot.version}</version> <scope>test</scope></dependency>
测试类需要添加注解@SpringBootTest
、@RunWith
,如下:
packagetk.fishfish.easyjava.log;importorg.junit.Test;importorg.junit.runner.RunWith;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.boot.test.context.SpringBootTest;importorg.springframework.test.context.junit4.SpringRunner;/** * 测试日志 * * @author 奔波儿灞 * @since 1.0 */@SpringBootTest@RunWith(SpringRunner.class)publicclassMyLogTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(MyLogTest.class);@Testpublicvoidinfo() {LOG.info("info示例,参数:{}","值"); }}
具体代码见:
tk.fishfish.easyjava.logMyLogTest
:测试日志
数据库连接池有很多成熟的产品,譬如c3p0
、dbcp
,国内的druid
等。
但是这些数据库连接池这里不做介绍,这里介绍HikariCP。
用MySQL
的朋友们请不要使用utf8
,请使用utf8mb4
。
推荐阅读:
警告:后文默认使用了utf8mb4
编码。
依赖:
<dependency> <groupId>com.zaxxer</groupId> <artifactId>HikariCP</artifactId> <version>3.3.1</version></dependency><!-- MySQL驱动(可选)--><dependency> <groupId>mysql</groupId> <artifactId>mysql-connector-java</artifactId> <version>5.1.46</version></dependency>
推荐阅读:
测试:
packagetk.fishfish.easyjava.datasource;importcom.zaxxer.hikari.HikariConfig;importcom.zaxxer.hikari.HikariDataSource;importorg.junit.Before;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjavax.sql.DataSource;importjava.sql.Connection;importjava.sql.PreparedStatement;importjava.sql.ResultSet;importjava.sql.SQLException;/** * Hikari测试 * * @author 奔波儿灞 * @since 1.0 */publicclassHikariTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(HikariTest.class);privateDataSourcedataSource;@Beforepublicvoidsetup() {HikariConfigconfig =newHikariConfig();config.setJdbcUrl("jdbc:mysql://localhost:3306/easyjava?useSSL=false&useUnicode=true&characterEncoding=utf-8");config.setUsername("root");config.setPassword("123456");config.setMinimumIdle(8);config.setMaximumPoolSize(32);// MySQL数据库编码设置为utf8mb4config.addDataSourceProperty("connectionInitSql","set names utf8mb4;");// MySQL推荐配置config.addDataSourceProperty("cachePrepStmts",true);config.addDataSourceProperty("prepStmtCacheSize",250);config.addDataSourceProperty("prepStmtCacheSqlLimit",2048);config.addDataSourceProperty("useServerPrepStmts",true);config.addDataSourceProperty("useLocalSessionState",true);config.addDataSourceProperty("rewriteBatchedStatements",true);config.addDataSourceProperty("cacheResultSetMetadata",true);config.addDataSourceProperty("cacheServerConfiguration",true);config.addDataSourceProperty("elideSetAutoCommits",true);config.addDataSourceProperty("maintainTimeStats",false);dataSource =newHikariDataSource(config); }@Testpublicvoidrun() {Connectionconnection =null;PreparedStatementps =null;ResultSetrs =null;try {connection =dataSource.getConnection();ps =connection.prepareStatement("show tables");rs =ps.executeQuery();while (rs.next()) {Stringtable =rs.getString(1);LOG.info("table: {}",table); } }catch (SQLExceptione) {LOG.error("数据库异常",e); }finally {// 释放if (rs !=null) {try {rs.close(); }catch (SQLExceptione) {LOG.error("数据库ResultSet关闭异常",e); } }if (ps !=null) {try {ps.close(); }catch (SQLExceptione) {LOG.error("数据库PreparedStatement关闭异常",e); } }if (connection !=null) {try {connection.close(); }catch (SQLExceptione) {LOG.error("数据库连接关闭异常",e); } } } }}
具体代码见:
tk.fishfish.easyjava.datasource.HikariTest
:Hikari测试
其实从spring-boot-2.0
开始,HikariCP
就已经作为默认的数据库连接池了。
如果你使用spring-boot-starter-jdbc
或spring-boot-starter-data-jpa
,会自动添加对HikariCP
的依赖。
依赖:
<dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-jdbc</artifactId> <version>${spring.boot.version}</version></dependency>
配置:
spring:datasource:driver-class-name:com.mysql.jdbc.Driverurl:jdbc:mysql://localhost:3306/easyjava?useSSL=false&useUnicode=true&characterEncoding=utf-8username:rootpassword:123456# 参数配置hikari:pool-name:hikariPoolminimum-idle:8maximum-pool-size:50# 初始化时设置编码connection-init-sql:set names utf8mb4;cache-prep-stmts:trueprep-stmt-cache-size:250prep-stmt-cache-sql-limit:2048use-server-prep-stmts:trueuse-local-session-state:truerewrite-batched-statements:truecache-result-set-metadata:truecache-server-configuration:trueelide-set-auto-commits:truemaintain-time-stats:false
配置之后,会默认配置好数据源以及jdbcTemplate
。
下面是jdbcTemplate
使用:
packagetk.fishfish.easyjava.datasource;importorg.junit.Test;importorg.junit.runner.RunWith;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.beans.factory.annotation.Autowired;importorg.springframework.boot.test.context.SpringBootTest;importorg.springframework.jdbc.core.JdbcTemplate;importorg.springframework.test.context.junit4.SpringRunner;importjava.util.List;importjava.util.Map;/** * 测试jdbcTemplate * * @author 奔波儿灞 * @since 1.0 */@SpringBootTest@RunWith(SpringRunner.class)publicclassJdbcTemplateTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(JdbcTemplateTest.class);@AutowiredprivateJdbcTemplatejdbcTemplate;@Testpublicvoidrun() {List<Map<String,Object>>tables =jdbcTemplate.queryForList("show tables");LOG.info("tables: {}",tables); }}
测试代码见:
tk.fishfish.easyjava.datasource.JdbcTemplateTest
:测试jdbcTemplate
依赖:
<dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-data-jpa</artifactId> <version>${spring.boot.version}</version></dependency>
配置:
spring:datasource:driver-class-name:com.mysql.jdbc.Driverurl:jdbc:mysql://localhost:3306/easyjava?useSSL=false&useUnicode=true&characterEncoding=utf-8username:rootpassword:123456# 参数配置hikari:pool-name:hikariPoolminimum-idle:8maximum-pool-size:50# 初始化时设置编码connection-init-sql:set names utf8mb4;cache-prep-stmts:trueprep-stmt-cache-size:250prep-stmt-cache-sql-limit:2048use-server-prep-stmts:trueuse-local-session-state:truerewrite-batched-statements:truecache-result-set-metadata:truecache-server-configuration:trueelide-set-auto-commits:truemaintain-time-stats:falsejpa:database-platform:org.hibernate.dialect.MySQL5Dialectshow-sql:truehibernate:ddl-auto:update
具体的测试代码见:
tk.fishfish.easyjava.datasource.Demo
:实体tk.fishfish.easyjava.datasource.repository.DemoRepository
:DAOtk.fishfish.easyjava.datasource.DemoRepositoryTest
:DAO测试
建议阅读:
关于ORM,目前使用比较多的有:
spring-boot-starter-data-jpa
Hibernate
Mybatis
spring-data-starter-jdbc
- 原生
JDBC
其中,封装程度、面向对象由深至浅。
但是现在已经是2019年了,Hibernate
使用已经很少了,虽然spring-boot-starter-data-jpa
底层使用了Hibernate
,但是不建议直接使用Hibernate
,使用更上层的spring-boot-starter-data-jpa
比较好,文档多,更新快。原生JDBC
编程也比较少了,建议使用上层一点的spring-data-starter-jdbc
。
因此,对于业务单一、简单,没有太多的复杂关联,推荐使用spring-boot-starter-data-jpa
。对于业务复杂,有比较多的业务关联,推荐使用Mybatis
,然后配合一些插件来使用。
关于Mybatis
的文档也比较多,这里推荐几个好用插件:
其中PageHelper
、Mappe
插件一般配合使用,分别用于分页、通过Mapper。mybatis-plus
包含上述2个插件的功能,还有其他附加功能。
大家灵活选择合适的插件即可。
这里推荐几个redis
客户端:
jedis
:老牌客户端lettuce
:新生客户端,提供异步操作,redis
各种模式redisson
:不仅仅是客户端了,还提供基于redis
的一系列生态(分布式锁、集合等)
从spring-boot2.x
开始,已经将lettuce
作为默认的redis
操作的底层实现了。
依赖:
<dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-data-redis</artifactId> <version>${spring.boot.version}</version></dependency>
默认情况下,spring-boot
会为我们配置StringRedisTemplate
和RedisTemplate<Object, Object>
这两个bean。
配置(Redis单实例):
spring:redis:database:1host:127.0.0.1port:6379password:timeout:3000lettuce:port:max-active:16max-idle:8min-idle:4max-wait:50ms
测试:
packagetk.fishfish.easyjava.redis;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.beans.factory.annotation.Autowired;importorg.springframework.data.redis.core.BoundValueOperations;importorg.springframework.data.redis.core.StringRedisTemplate;importjava.util.concurrent.TimeUnit;/** * 测试RedisTemplate * * @author 奔波儿灞 * @since 1.0 */publicclassRedisTemplateTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(RedisTemplateTest.class);@AutowiredprivateStringRedisTemplateredisTemplate;@Testpublicvoidrun() {finalStringkey ="esayjava";finalStringvalue ="易学Java";BoundValueOperations<String,String>ops =redisTemplate.boundValueOps(key);ops.set(value);// 1分钟过期ops.expire(1,TimeUnit.MINUTES);Stringv =ops.get();LOG.info("v: {}",v); }}
测试代码见:
tk.fishfish.easyjava.redis.RedisTemplateTest
:测试RedisTemplate
接下来,你需要学习如何配置RedisTemplate
序列化,集成cache
等高级功能。
redis4.0
以前,删除bigkey
会导致阻塞(redis单线程模型)。
因此,string
类型控制在10KB以内,hash
、list
、set
、zset
元素个数不要超过5000,拒绝bigKey。
提示:redis 4.0已经支持key的异步删除,欢迎使用
publicvoiddelBigHash(Stringhost,intport,Stringpassword,StringbigHashKey) {Jedisjedis =newJedis(host,port);if (password !=null && !"".equals(password)) {jedis.auth(password); }ScanParamsscanParams =newScanParams().count(100);Stringcursor ="0";do {ScanResult<Entry<String,String>>scanResult =jedis.hscan(bigHashKey,cursor,scanParams);List<Entry<String,String>>entryList =scanResult.getResult();if (entryList !=null && !entryList.isEmpty()) {for (Entry<String,String>entry :entryList) {jedis.hdel(bigHashKey,entry.getKey()); } }cursor =scanResult.getStringCursor(); }while (!"0".equals(cursor));//删除bigkeyjedis.del(bigHashKey);}
publicvoiddelBigList(Stringhost,intport,Stringpassword,StringbigListKey) {Jedisjedis =newJedis(host,port);if (password !=null && !"".equals(password)) {jedis.auth(password); }longllen =jedis.llen(bigListKey);intcounter =0;intleft =100;while (counter <llen) {//每次从左侧截掉100个jedis.ltrim(bigListKey,left,llen);counter +=left; }//最终删除keyjedis.del(bigListKey);}
publicvoiddelBigSet(Stringhost,intport,Stringpassword,StringbigSetKey) {Jedisjedis =newJedis(host,port);if (password !=null && !"".equals(password)) {jedis.auth(password); }ScanParamsscanParams =newScanParams().count(100);Stringcursor ="0";do {ScanResult<String>scanResult =jedis.sscan(bigSetKey,cursor,scanParams);List<String>memberList =scanResult.getResult();if (memberList !=null && !memberList.isEmpty()) {for (Stringmember :memberList) {jedis.srem(bigSetKey,member); } }cursor =scanResult.getStringCursor(); }while (!"0".equals(cursor));// 删除bigkeyjedis.del(bigSetKey);}
publicvoiddelBigZset(Stringhost,intport,Stringpassword,StringbigZsetKey) {Jedisjedis =newJedis(host,port);if (password !=null && !"".equals(password)) {jedis.auth(password); }ScanParamsscanParams =newScanParams().count(100);Stringcursor ="0";do {ScanResult<Tuple>scanResult =jedis.zscan(bigZsetKey,cursor,scanParams);List<Tuple>tupleList =scanResult.getResult();if (tupleList !=null && !tupleList.isEmpty()) {for (Tupletuple :tupleList) {jedis.zrem(bigZsetKey,tuple.getElement()); } }cursor =scanResult.getStringCursor(); }while (!"0".equals(cursor));// 删除bigkeyjedis.del(bigZsetKey);}
建议阅读:
先看一段代码:
publicbooleantryAcquire() {Jedisjedis =null;try {jedis =pool.getResource();// 获取当前剩余的凭据数Longcurrent =Long.valueOf(jedis.get(key));if (current >0) {// 凭据数大于0,则获取成功,减一jedis.incr(key);returntrue; }returnfalse; }catch (JedisExceptione) {LOG.error("tryAcquire error",e);returnfalse; }finally {returnResource(jedis); }}
上面的例子是获取redis
凭据,如果凭据数大于0,则获取成功,减一;否则获取失败。
感觉是没问题的,但是稍微接触过多线程,就会发现,get(key)
、incr(key)
这两个操作组成的代码块不是原子性。
为了保证这个操作是原子性,我们可以用lua
脚本将get
、incr
命令封装成原子性。
/** * 基于redis lua实现Semaphore * * @author xuan * @since 1.0.0 */publicclassRedisSemaphoreimplementsSemaphore {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(Semaphore.class);/** * redis默认存储的key */privatestaticfinalStringDEFAULT_KEY ="rateLimit:semaphore";/** * lua执行脚本,如果大于0,则减一,返回1,代表获取成功 */privatestaticfinalStringSCRIPT_LIMIT ="local key = KEYS[1] " +"local current = tonumber(redis.call('get', key)) " +"local res = 0 " +"if current > 0 then " +" redis.call('decr', key) " +" res = 1 " +"end " +"return res ";/** * Redis连接池 */privatefinalPool<Jedis>pool;/** * redis存储的key */privatefinalStringkey;/** * 凭据限制的数目 */privatefinalLonglimits;publicRedisSemaphore(Pool<Jedis>pool,Longlimits) {this(pool,DEFAULT_KEY,limits); }publicRedisSemaphore(Pool<Jedis>pool,Stringkey,Longlimits) {this.pool =pool;this.key =key;this.limits =limits;setup(); }/** * 尝试获取凭据,获取不到凭据不等待,直接返回 * * @return 获取到凭据返回true,否则返回false */@OverridepublicbooleantryAcquire() {Jedisjedis =null;try {jedis =pool.getResource();Longres = (Long)jedis.eval(SCRIPT_LIMIT,Collections.singletonList(key),Collections.<String>emptyList());returnres >0; }catch (JedisExceptione) {LOG.error("tryAcquire error",e);returnfalse; }finally {returnResource(jedis); } }/** * 释放获取到的凭据 */@Overridepublicvoidrelease() {Jedisjedis =null;try {jedis =pool.getResource();jedis.incr(key); }catch (JedisExceptione) {LOG.error("release error",e); }finally {returnResource(jedis); } }privatevoidsetup() {Jedisjedis =null;try {jedis =pool.getResource();jedis.del(key);jedis.incrBy(key,limits); }finally {returnResource(jedis); } }privatevoidreturnResource(Jedisjedis) {if (jedis !=null) {jedis.close(); } }}
建议搜索下lua
脚本的相关知识,比如jedis
怎么使用、spring
中怎么使用等等。
分布式锁实现方案比较多:
Redis
分布式锁Zookeeper
分布式锁
建议大家搜索相关资料学习下原理。
推荐使用redisson
封装的分布式锁。
Guava
中令牌桶实现Redis
实现
建议大家搜索相关资料学习下原理。
推荐使用redisson
封装的限流器
如果你在寻求通过redis实现某些功能(比如布隆过滤器、阻塞队列等),一定要看看redisson
,也许你忙活了半天要实现的功能,人家已经实现了呢?(关键是实现的更健壮哟)。
这里不推荐直接使用poi
,推荐使用阿里巴巴封装的easyexcel组件。
关于easyexcel
的使用,这里就不做介绍,直接看官方文档即可。
###类型转换
easyexcel
通过com.alibaba.excel.converters.DefaultConverterLoader
注册一些默认的类型转化
当有个字段类型为java.sql.Timestamp
时,是无法正常转换的,会报错(com.alibaba.excel.exception.ExcelDataConvertException: Can not find 'Converter' support class Timestamp.
)。
此时,我们需要自定义类型转换。
packagetk.fishfish.easyjava.excel;importcom.alibaba.excel.converters.Converter;importcom.alibaba.excel.enums.CellDataTypeEnum;importcom.alibaba.excel.metadata.CellData;importcom.alibaba.excel.metadata.GlobalConfiguration;importcom.alibaba.excel.metadata.property.ExcelContentProperty;importcom.alibaba.excel.util.DateUtils;importjava.sql.Timestamp;importjava.util.Date;/** * Timestamp转换 * * @author 奔波儿灞 * @see com.alibaba.excel.converters.date.DateStringConverter * @since 1.0 */publicclassTimestampConverterimplementsConverter<Timestamp> {@OverridepublicClasssupportJavaTypeKey() {returnTimestamp.class; }@OverridepublicCellDataTypeEnumsupportExcelTypeKey() {returnCellDataTypeEnum.STRING; }@OverridepublicTimestampconvertToJavaData(CellDatacellData,ExcelContentPropertycontentProperty,GlobalConfigurationconfiguration)throwsException {if (contentProperty ==null ||contentProperty.getDateTimeFormatProperty() ==null) {Datedate =DateUtils.parseDate(cellData.getStringValue(),null);returnnewTimestamp(date.getTime()); }else {Datedate =DateUtils.parseDate(cellData.getStringValue(),contentProperty.getDateTimeFormatProperty().getFormat());returnnewTimestamp(date.getTime()); } }@OverridepublicCellDataconvertToExcelData(Timestampvalue,ExcelContentPropertycontentProperty,GlobalConfigurationconfiguration) {if (contentProperty ==null ||contentProperty.getDateTimeFormatProperty() ==null) {returnnewCellData(DateUtils.format(value,null)); }else {returnnewCellData(DateUtils.format(value,contentProperty.getDateTimeFormatProperty().getFormat())); } }}
这里参考官方的com.alibaba.excel.converters.date.DateStringConverter
,写一个自定义的转换。
接下来注册到easyexcel
:
packagetk.fishfish.easyjava.excel;importcom.alibaba.excel.EasyExcel;importorg.junit.Test;importjava.sql.Timestamp;importjava.util.ArrayList;importjava.util.Collections;importjava.util.Date;importjava.util.HashMap;importjava.util.List;importjava.util.Map;importjava.util.stream.Collectors;/** * easyexcel测试 * * @author 奔波儿灞 * @since 1.0 */publicclassEasyExcelTest {@TestpublicvoidcustomConverter() {List<Map<String,Object>>data =newArrayList<>();Map<String,Object>row =newHashMap<>();row.put("name","名字");row.put("age",10);// 官方无法转换Timestamprow.put("time",newTimestamp(System.currentTimeMillis()));row.put("date",newDate());row.put("sex",1);data.add(row);EasyExcel.write("demo.xlsx") .head(head(data))// 注册自定义类型转换 .registerConverter(newTimestampConverter()) .sheet("data") .doWrite(data(data)); }privateList<List<String>>head(List<Map<String,Object>>data) {returndata.stream() .findFirst() .map(Map::keySet) .map(set ->set.stream().map(Collections::singletonList).collect(Collectors.toList())) .orElseGet(Collections::emptyList); }privateList<List<Object>>data(List<Map<String,Object>>data) {returndata.stream() .map(Map::values) .map(ArrayList::new) .collect(Collectors.toList()); }}
参考官方的测试代码即可。
网上找找资料,了解双亲委托模型。
记住:
加载class时,一直委托给父
ClassLoader
加载。直到最终无法加载,才自己加载。不同的
ClassLoader
可以加载同一class(全类名相同),生成不同实例。但是,这两个class实例生成的对象不能强转
继承java.lang.ClassLoader
对象,覆盖findClass
方法。一般的,为了加载本地目录的某个jar,继承java.net.URLClassLoader
即可。
packagetk.fishfish.easyjava.jvm;importjava.net.URL;importjava.net.URLClassLoader;/** * 自定义ClassLoader * * @author 奔波儿灞 * @since 1.0 */publicclassMyClassLoaderextendsURLClassLoader {publicMyClassLoader(URL[]urls,ClassLoaderparent) {super(urls,parent); }}
这里采用自定义的tk.fishfish.easyjava.jvm.MyClassLoader
来加载本地打包的jar,然后比较加载的Class与系统加载的同一类的Class是否相同。
packagetk.fishfish.easyjava.jvm;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importtk.fishfish.easyjava.Application;importjava.io.File;importjava.net.MalformedURLException;importjava.net.URL;/** * 测试自定义ClassLoader * * @author 奔波儿灞 * @since 1.0 */publicclassMyClassLoaderTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(MyClassLoaderTest.class);@TestpublicvoidloadClass()throwsMalformedURLException,ClassNotFoundException {Filejar =newFile("target/easy-java-1.0.0-SNAPSHOT.jar.original");if (jar.isFile()) {URLtargetUrl =jar.toURI().toURL();// 加载,不指定parent,则parent默认为系统类加载器MyClassLoaderclassLoader =newMyClassLoader(newURL[]{targetUrl},null);Class<?>clazz =Class.forName("tk.fishfish.easyjava.Application",true,classLoader);LOG.info("equals: {}",clazz.equals(Application.class)); }else {thrownewRuntimeException("must be jar"); } }}
显然,tk.fishfish.easyjava.Application
类由不同的ClassLoader加载,他们的Class调用equals方法比较是不同的。
这里为什么不用target/easy-java-1.0.0-SNAPSHOT.jar
测试?因为,spring boot打包有点特殊,是无法加载到tk.fishfish.easyjava.Application
的。
原因就是,spring boot也是自定义ClassLoader来加载jar咯。可以参考该博客,了解一下。
有了自定义ClassLoader,那么可以实现热加载、相同类名的隔离(全类名相同的类,但是代码不同)等。
SPI
全称Service Provider Interface
,是Java
提供的一套用来被第三方实现或者扩展的API
,它可以用来启用框架扩展和替换组件。
比较常见的例子:
- 数据库驱动(Driver)加载接口实现类的加载
SLF4J
加载不同提供商的日志实现类- Spring中大量使用了SPI,比如:对servlet 3.0规范对
ServletContainerInitializer
的实现、自动类型转换Type Conversion SPI(Converter SPI、Formatter SPI)等 - Dubbo、Motan等RPC框架中也大量使用SPI的方式实现框架的扩展,不过它对Java提供的原生SPI做了封装,允许用户扩展实现Filter接口
- 当服务提供者提供了接口的一种具体实现后,在jar包的
META-INF/services
目录下创建一个以"接口全限定名"为命名的文件,内容为实现类的全限定名 - 接口实现类所在的jar包放在主程序的classpath中
- 主程序通过
java.util.ServiceLoder
动态装载实现模块,它通过扫描META-INF/services
目录下的配置文件找到实现类的全限定名,把类加载到JVM - SPI的实现类必须携带一个不带参数的构造方法
比如,我们查找下java.sql.Driver
接口的提供类有哪些:
packagetk.fishfish.easyjava.jvm;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.sql.Driver;importjava.util.ServiceLoader;/** * SPI机制 * * @author 奔波儿灞 * @since 1.0 */publicclassSpiTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(SpiTest.class);@Testpublicvoidloader() {ServiceLoader<Driver>driverServiceLoader =ServiceLoader.load(Driver.class);for (Driverdriver :driverServiceLoader) {LOG.info("driver: {}",driver.getClass().getName()); } }}
可以看到,找到了MySQL的驱动实现。
我们看MySQL的驱动包结构,也能发现在META-INF/services
目录下有个java.sql.Driver
文件名的文件,里面是实现类的全类名。
在自己开发一些框架的时候,如果提供插件。可使用SPI机制,动态查找到实现类,从而可插拔,而又不需要大量的配置或耦合spring的依赖注入。
java util concurrent包
对于Java1.7,是基于segments分段进行实现。漫画:什么是ConcurrentHashMap
Java1.8则已经抛弃了segment分段锁机制,利用CAS+Synchronized来保证并发更新的安全,底层依然采用数组+链表+红黑树的存储结构。
对于put操作:
finalVputVal(Kkey,Vvalue,booleanonlyIfAbsent) {if (key ==null ||value ==null)thrownewNullPointerException();inthash =spread(key.hashCode());intbinCount =0;for (Node<K,V>[]tab =table;;) {Node<K,V>f;intn,i,fh;if (tab ==null || (n =tab.length) ==0)tab =initTable();elseif ((f =tabAt(tab,i = (n -1) &hash)) ==null) {if (casTabAt(tab,i,null,newNode<K,V>(hash,key,value,null)))break;// no lock when adding to empty bin }elseif ((fh =f.hash) ==MOVED)tab =helpTransfer(tab,f);else {VoldVal =null;synchronized (f) {if (tabAt(tab,i) ==f) {if (fh >=0) {binCount =1;for (Node<K,V>e =f;; ++binCount) {Kek;if (e.hash ==hash && ((ek =e.key) ==key || (ek !=null &&key.equals(ek)))) {oldVal =e.val;if (!onlyIfAbsent)e.val =value;break; }Node<K,V>pred =e;if ((e =e.next) ==null) {pred.next =newNode<K,V>(hash,key,value,null);break; } } }elseif (finstanceofTreeBin) {Node<K,V>p;binCount =2;if ((p = ((TreeBin<K,V>)f).putTreeVal(hash,key,value)) !=null) {oldVal =p.val;if (!onlyIfAbsent)p.val =value; } } } }if (binCount !=0) {if (binCount >=TREEIFY_THRESHOLD)treeifyBin(tab,i);if (oldVal !=null)returnoldVal;break; } } }addCount(1L,binCount);returnnull;}
对于get操作:
publicVget(Objectkey) {Node<K,V>[]tab;Node<K,V>e,p;intn,eh;Kek;inth =spread(key.hashCode());if ((tab =table) !=null && (n =tab.length) >0 && (e =tabAt(tab, (n -1) &h)) !=null) {if ((eh =e.hash) ==h) {if ((ek =e.key) ==key || (ek !=null &&key.equals(ek)))returne.val; }elseif (eh <0)return (p =e.find(h,key)) !=null ?p.val :null;while ((e =e.next) !=null) {if (e.hash ==h && ((ek =e.key) ==key || (ek !=null &&key.equals(ek))))returne.val; } }returnnull;}
可以看到get操作全程没有加锁,由于table、node均使用了volatile修饰,因此保证了可见性。所以get是弱一致性。
ConcurrentSkipListMap
是基于SkipList(跳表)来实现Map,利用CAS来保证并发更新的安全。
采用分层索引的结构,以空间换时间。
ConcurrentLinkedQueue
是一个基于链表的无界线程安全的非阻塞队列,按照先进先出原则(FIFO)对元素进行排序。新元素从队列尾部插入,而获取队列元素,则需要从队列头部获取。
从源码中能看出来,底层是基于CAS进行并发控制的。
publicbooleanoffer(Ee) {checkNotNull(e);finalNode<E>newNode =newNode<E>(e);for (Node<E>t =tail,p =t;;) {Node<E>q =p.next;if (q ==null) {// p is last nodeif (p.casNext(null,newNode)) {// Successful CAS is the linearization point// for e to become an element of this queue,// and for newNode to become "live".if (p !=t)// hop two nodes at a timecasTail(t,newNode);// Failure is OK.returntrue; }// Lost CAS race to another thread; re-read next }elseif (p ==q)// We have fallen off list. If tail is unchanged, it// will also be off-list, in which case we need to// jump to head, from which all live nodes are always// reachable. Else the new tail is a better bet.p = (t != (t =tail)) ?t :head;else// Check for tail updates after two hops.p = (p !=t &&t != (t =tail)) ?t :q; }}publicEpoll() {restartFromHead:for (;;) {for (Node<E>h =head,p =h,q;;) {Eitem =p.item;if (item !=null &&p.casItem(item,null)) {// Successful CAS is the linearization point// for item to be removed from this queue.if (p !=h)// hop two nodes at a timeupdateHead(h, ((q =p.next) !=null) ?q :p);returnitem; }elseif ((q =p.next) ==null) {updateHead(h,p);returnnull; }elseif (p ==q)continuerestartFromHead;elsep =q; } }}
CopyOnWriteArrayList
是ArrayList
的线程安全版本,在有写操作的时候会copy一份数据,然后写完再设置成新的数据。适用于读多写少的并发场景。
/** The lock protecting all mutators */finaltransientReentrantLocklock =newReentrantLock();/** The array, accessed only via getArray/setArray. */privatetransientvolatileObject[]array;
- spring cloud netflix eureka client存储服务列表
写操作:
- 获取锁
- 复制数组,长度+1,并赋值
- 设置新array的引用
- 释放锁
publicbooleanadd(Ee) {finalReentrantLocklock =this.lock;lock.lock();try {Object[]elements =getArray();intlen =elements.length;Object[]newElements =Arrays.copyOf(elements,len +1);newElements[len] =e;setArray(newElements);returntrue; }finally {lock.unlock(); }}
读操作:
- 直接根据下标获取array对应的值
privateEget(Object[]a,intindex) {return (E)a[index];}/** * {@inheritDoc} * * @throws IndexOutOfBoundsException {@inheritDoc} */publicEget(intindex) {returnget(getArray(),index);}
迭代:
- 获取当前array的引用。如果后续array变化了(因为每次写操作都会有新的拷贝),迭代器是感知不到的。
publicIterator<E>iterator() {returnnewCOWIterator<E>(getArray(),0);}
SynchronousQueue
是无界的,是一种无缓冲的等待队列,但是由于该Queue本身的特性,在某次添加元素后必须等待其他线程取走后才能继续添加;可以认为SynchronousQueue
是一个缓存值为1的阻塞队列,但是isEmpty()方法永远返回是true,remainingCapacity() 方法永远返回是0,remove()和removeAll() 方法永远返回是false,iterator()方法永远返回空,peek()方法永远返回null。
可以理解为"配对"队列。
内部没有存储
阻塞队列
发送或者消费线程会阻塞,只要有一对消费和发送线程匹配上,才同时退出。
配对有公平模式和非公平模式(默认)
公平模式用队列实现 ,每次从队列head开始匹配。(FIFO)
非公平模式用栈实现,每次从栈顶开始匹配。(LIFO)
publicSynchronousQueue() {this(false);}publicSynchronousQueue(booleanfair) {transferer =fair ?newTransferQueue<E>() :newTransferStack<E>();}
主要应用在线程池:
packagetk.fishfish.easyjava.concurrent;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.SynchronousQueue;/** * SynchronousQueue测试 * * @author 奔波儿灞 * @version 1.0 */publicclassSynchronousQueueTest {privatefinalLoggerlogger =LoggerFactory.getLogger(SynchronousQueueTest.class);@Testpublicvoidrun()throwsInterruptedException {SynchronousQueue<String>queue =newSynchronousQueue<>(true);// 生产者newThread(newRunnable() {@Overridepublicvoidrun() {try {logger.info("PUT A start");queue.put("A");logger.info("PUT A end"); }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } }).start();Thread.sleep(1000);// 生产者newThread(newRunnable() {@Overridepublicvoidrun() {try {logger.info("PUT B start");queue.put("B");logger.info("PUT B end"); }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } }).start();Thread.sleep(1000);// taketry {logger.info("TAKE start");Stringitem =queue.take();logger.info("TAKE end: {}",item); }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); }Thread.sleep(1000); }}
在公平模式下输出:
18:19:14.726 [Thread-0] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - PUT A start18:19:15.729 [Thread-1] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - PUT B start18:19:16.733 [main] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - TAKE start18:19:16.733 [Thread-0] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - PUT A end18:19:16.733 [main] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - TAKE end: A
此时,获取的是A,唤醒的也是A。(队列的特点)
在不公平模式下输出:
18:19:46.784 [Thread-0] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - PUT A start18:19:47.787 [Thread-1] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - PUT B start18:19:48.789 [main] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - TAKE start18:19:48.790 [Thread-1] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - PUT B end18:19:48.790 [main] INFO tk.fishfish.easyjava.concurrent.SynchronousQueueTest - TAKE end: B
此时,获取的是B,唤醒的也是B。(栈的特点)
LinkedBlockingQueue
内部由单链表实现,只能从head取元素,从tail添加元素。添加元素和获取元素都有独立的锁,也就是说LinkedBlockingQueue
是读写分离的,读写操作可以并行执行。LinkedBlockingQueue
采用可重入锁(**ReentrantLock)**来保证在并发情况下的线程安全。
/** Lock held by take, poll, etc */privatefinalReentrantLocktakeLock =newReentrantLock();/** Wait queue for waiting takes */privatefinalConditionnotEmpty =takeLock.newCondition();/** Lock held by put, offer, etc */privatefinalReentrantLockputLock =newReentrantLock();/** Wait queue for waiting puts */privatefinalConditionnotFull =putLock.newCondition();
取数据
take():当队列为空时阻塞
poll():弹出队顶元素,队列为空时,返回空
peek():和poll类似,返回队顶元素,但顶元素不弹出。队列为空时返回null
remove(Object o):移除某个元素,队列为空时抛出异常。成功移除返回true
添加数据
put():队列满时阻塞
offer():队列满时返回false
一般用于生产者消费者模型:
packagetk.fishfish.easyjava.concurrent;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.ArrayList;importjava.util.List;importjava.util.UUID;importjava.util.concurrent.LinkedBlockingQueue;importjava.util.concurrent.TimeUnit;/** * LinkedBlockingQueue测试 * * @author 奔波儿灞 * @version 1.0 */publicclassLinkedBlockingQueueTest {privatefinalLoggerlogger =LoggerFactory.getLogger(LinkedBlockingQueueTest.class);@Testpublicvoidrun() {LinkedBlockingQueue<String>queue =newLinkedBlockingQueue<>(1024);// 生产者newThread(newRunnable() {@Overridepublicvoidrun() {while (true) {try {// 队列满了后等1s,如果仍然满了,则丢弃booleanoffer =queue.offer(UUID.randomUUID().toString(),1,TimeUnit.SECONDS);if (!offer) {logger.warn("queue full"); } }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } } }).start();// 消费者处理newThread(newRunnable() {@Overridepublicvoidrun() {while (true) {try {Stringitem =queue.poll(1,TimeUnit.SECONDS);if (item ==null) {logger.info("no item"); }else {// 模拟消费时常Thread.sleep(3000);logger.info("item: {}",item); } }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } } }).start();// 消费者处理,每次提取多个newThread(newRunnable() {privatefinalintcapacity =128;privatefinalList<String>buffer =newArrayList<>(capacity);@Overridepublicvoidrun() {while (true) {// 每次提取128个,数据不够时等待500msintsize =queue.drainTo(buffer,capacity);logger.info("drain size: {}",size);buffer.clear();try {Thread.sleep(10000);if (size <capacity) {Thread.sleep(500); } }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } } }).start();// main等待try {Thread.sleep(30_000); }catch (InterruptedExceptione) {logger.warn("main Thread interrupted",e); } }}
put过程:
put锁
判断队列是否满了,满了则等待
enqueue(node),在队尾增加元素
队列长度加1,此时如果队列还没有满,唤醒其他put堵塞队列
publicvoidput(Ee)throwsInterruptedException {if (e ==null)thrownewNullPointerException();// Note: convention in all put/take/etc is to preset local var// holding count negative to indicate failure unless set.intc = -1;Node<E>node =newNode<E>(e);finalReentrantLockputLock =this.putLock;finalAtomicIntegercount =this.count;putLock.lockInterruptibly();try {/* * Note that count is used in wait guard even though it is * not protected by lock. This works because count can * only decrease at this point (all other puts are shut * out by lock), and we (or some other waiting put) are * signalled if it ever changes from capacity. Similarly * for all other uses of count in other wait guards. */while (count.get() ==capacity) {notFull.await(); }enqueue(node);c =count.getAndIncrement();if (c +1 <capacity)notFull.signal(); }finally {putLock.unlock(); }if (c ==0)signalNotEmpty();}
take过程:
- take锁
- 没有数据,则阻塞;否则dequeue()从队列头部获取数据
- 判断队列长度是否大于1,大于则唤醒其他take阻塞队列
publicEtake()throwsInterruptedException {Ex;intc = -1;finalAtomicIntegercount =this.count;finalReentrantLocktakeLock =this.takeLock;takeLock.lockInterruptibly();try {while (count.get() ==0) {notEmpty.await(); }x =dequeue();c =count.getAndDecrement();if (c >1)notEmpty.signal(); }finally {takeLock.unlock(); }if (c ==capacity)signalNotFull();returnx;}
结合LinkedBlockingQueue
和SynchronousQueue
的特点实现。相比较SynchronousQueue
多了一个可以存储的队列,相比较LinkedBlockingQueue
多了直接传递元素,以及无阻塞。
底层使用链表存储数据,基于CAS进行并发控制,并能直接从head交换数据。
/** * Inserts the specified element at the tail of this queue. * As the queue is unbounded, this method will never block. * * @throws NullPointerException if the specified element is null */publicvoidput(Ee) {xfer(e,true,ASYNC,0);}
PriorityBlockingQueue
是优先级阻塞队列,底层采用数组存储数据,并根据compareTo接口进行排序,利用可重入锁(**ReentrantLock)**来保证在并发情况下的线程安全。
扩容。
添加元素时,检测容量已满,则进行扩容。
publicbooleanoffer(Ee) {if (e ==null)thrownewNullPointerException();finalReentrantLocklock =this.lock;lock.lock();intn,cap;Object[]array;while ((n =size) >= (cap = (array =queue).length))tryGrow(array,cap);try {Comparator<?superE>cmp =comparator;if (cmp ==null)siftUpComparable(n,e,array);elsesiftUpUsingComparator(n,e,array,cmp);size =n +1;notEmpty.signal(); }finally {lock.unlock(); }returntrue;}
当容量小于64时,每次扩容oldCap + 2;否则扩容oldCap >> 1(原始容量的一半)。
privatevoidtryGrow(Object[]array,intoldCap) {lock.unlock();// must release and then re-acquire main lockObject[]newArray =null;if (allocationSpinLock ==0 &&UNSAFE.compareAndSwapInt(this,allocationSpinLockOffset,0,1)) {try {intnewCap =oldCap + ((oldCap <64) ? (oldCap +2) :// grow faster if small (oldCap >>1));if (newCap -MAX_ARRAY_SIZE >0) {// possible overflowintminCap =oldCap +1;if (minCap <0 ||minCap >MAX_ARRAY_SIZE)thrownewOutOfMemoryError();newCap =MAX_ARRAY_SIZE; }if (newCap >oldCap &&queue ==array)newArray =newObject[newCap]; }finally {allocationSpinLock =0; } }if (newArray ==null)// back off if another thread is allocatingThread.yield();lock.lock();if (newArray !=null &&queue ==array) {queue =newArray;System.arraycopy(array,0,newArray,0,oldCap); }}
DelayQueue
是BlockingQueue
的一种,所以它是线程安全的,DelayQueue
的特点就是插入Queue中的数据可以按照自定义的delay时间进行排序。只有delay时间小于0的元素才能够被取出。
DelayQueue
一般用于生产者消费者模式。
首先定义Delayed
对象:
packagetk.fishfish.easyjava.concurrent;importjava.util.concurrent.Delayed;importjava.util.concurrent.TimeUnit;/** * Delayed对象 * * @author 奔波儿灞 * @version 1.0 */publicclassDelayData<T>implementsDelayed {/** * 数据 */privatefinalTitem;/** * 过期时间 */privatefinallongexpireTimestamp;publicDelayData(Titem,longexpireTimestamp) {this.item =item;this.expireTimestamp =expireTimestamp; }@OverridepubliclonggetDelay(TimeUnitunit) {longdiffTime =expireTimestamp -System.currentTimeMillis();returnunit.convert(diffTime,TimeUnit.MILLISECONDS); }@OverridepublicintcompareTo(Delayedo) {return (int) (this.expireTimestamp - ((DelayData<T>)o).getExpireTimestamp()); }publicTgetItem() {returnitem; }publiclonggetExpireTimestamp() {returnexpireTimestamp; }}
生产者消费者使用:
packagetk.fishfish.easyjava.concurrent;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.DelayQueue;importjava.util.concurrent.TimeUnit;/** * DelayQueue测试 * * @author 奔波儿灞 * @version 1.0 */publicclassDelayQueueTest {privatefinalLoggerlogger =LoggerFactory.getLogger(DelayQueueTest.class);@Testpublicvoidrun() {finalDelayQueue<DelayData<String>>queue =newDelayQueue<>();// 模拟生产者放入数据queue.add(newDelayData<>("ID=1",System.currentTimeMillis() +5_000));queue.add(newDelayData<>("ID=2",System.currentTimeMillis() +10_000));// 消费者处理newThread(newRunnable() {@Overridepublicvoidrun() {while (true) {try {DelayData<String>data =queue.poll(1,TimeUnit.SECONDS);if (data ==null) {logger.info("no item"); }else {logger.info("item: {}",data.getItem()); } }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } } }).start();// main等待try {Thread.sleep(30_000); }catch (InterruptedExceptione) {logger.warn("main Thread interrupted",e); } }}
输出:
15:47:49.531 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:50.536 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:51.542 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:52.546 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:53.529 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - item: ID=115:47:54.536 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:55.540 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:56.541 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:57.543 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:47:58.529 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - item: ID=215:47:59.534 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:48:00.538 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:48:01.540 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:48:02.545 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item15:48:03.548 [Thread-0] INFO tk.fishfish.easyjava.concurrent.DelayQueueTest - no item
取数据
take():当队列为空时阻塞
poll():弹出队顶元素,队列为空时,返回空
peek():和poll类似,返回队顶元素,但顶元素不弹出。队列为空时返回null
添加数据
put():队列满时阻塞
offer():队列满时返回false
DelayQueue
内部采用PriorityQueue
存储数据。
当调用add(offer)方法添加元素时,内部调用
PriorityQueue
的offer方法(底层要根据compareTo方法排序,一般把先过期的数据放前面)。当调用poll方法获取元素时,内部先调用
PriorityQueue
的peek方法获取第一个元素。如果存在,则调用getDelay方法判断是否达到时间,如果小于等于0则说明元素达到延迟时间,应该取出,此时调用内部PriorityQueue
的poll方法。publicEpoll() {finalReentrantLocklock =this.lock;lock.lock();try {Efirst =q.peek();if (first ==null ||first.getDelay(NANOSECONDS) >0)returnnull;elsereturnq.poll(); }finally {lock.unlock(); }}
Exchanger(交换者)是一个用于线程间协作的工具类。Exchanger用于进行线程间的数据交换。它提供一个同步点,在这个同步点两个线程可以交换彼此的数据。这两个线程通过exchange方法交换数据, 如果第一个线程先执行exchange方法,它会一直等待第二个线程也执行exchange,当两个线程都到达同步点时,这两个线程就可以交换数据,将本线程生产出来的数据传递给对方。
- 遗传算法,遗传算法里需要选出两个人作为交配对象,这时候会交换两人的数据,并使用交叉规则得出2个交配结果。
- 校对工作。比如我们需要将纸制银流通过人工的方式录入成电子银行流水,为了避免错误,采用AB岗两人进行录入,录入到Excel之后,系统需要加载这两个Excel,并对这两个Excel数据进行校对,看看是否录入的一致。
- 生产者消费者。生产者将数据交给消费者。
packagetk.fishfish.easyjava.concurrent;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.UUID;importjava.util.concurrent.Exchanger;/** * Exchanger测试 * * @author 奔波儿灞 * @version 1.0 */publicclassExchangerTest {privatefinalLoggerlogger =LoggerFactory.getLogger(ExchangerTest.class);@Testpublicvoidrun() {Exchanger<String>exchanger =newExchanger<>();// 生产者newThread(newRunnable() {@Overridepublicvoidrun() {while (true) {try {Stringdata =exchanger.exchange(UUID.randomUUID().toString());logger.info("投递给: {}",data); }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } } }).start();// 消费者newThread(newRunnable() {@Overridepublicvoidrun() {while (true) {try {Stringdata =exchanger.exchange("Consumer-1");logger.info("接收到: {}",data); }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); } } } }).start();// main等待try {Thread.sleep(30_000); }catch (InterruptedExceptione) {logger.warn("main Thread interrupted",e); } }}
输出:
11:00:17.657 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 投递给: Consumer-111:00:17.657 [Thread-1] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 接收到: 1673319b-c249-4ff4-be07-5be96b27ca0d11:00:18.664 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 投递给: Consumer-111:00:18.664 [Thread-1] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 接收到: c061c0af-ce0e-481e-ab0e-eac03552f42311:00:19.669 [Thread-1] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 接收到: 1ebe7482-a2be-4d85-8fd6-679834c2684911:00:19.669 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 投递给: Consumer-111:00:20.672 [Thread-1] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 接收到: 349a4bd9-40da-464d-9d12-5e90f9010e8511:00:20.675 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 投递给: Consumer-111:00:21.679 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 投递给: Consumer-111:00:21.679 [Thread-1] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 接收到: 026df8fa-a135-4949-af9f-fb3acb659cb511:00:22.683 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 投递给: Consumer-111:00:22.683 [Thread-1] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 接收到: 219d3258-3732-4652-b098-47665353d0b711:00:23.687 [Thread-1] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 接收到: 22ba56b4-78c9-4d6f-bb01-b4988d78219c11:00:23.687 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 投递给: Consumer-1
CyclicBarrier
的字面意思是可循环使用(Cyclic)的屏障(Barrier)。它要做的事情是,让一组线程到达一个屏障(也可以叫同步点)时被阻塞,直到最后一个线程到达屏障时,屏障才会开门,所有被屏障拦截的线程才会继续执行。CyclicBarrier默认的构造方法是CyclicBarrier(int parties),其参数表示屏障拦截的线程数量,每个线程调用await方法告诉CyclicBarrier
我已经到达了屏障,然后当前线程被阻塞。
- 任务组控制。所有线程都准备好之后,开始执行。
packagetk.fishfish.easyjava.concurrent;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.BrokenBarrierException;importjava.util.concurrent.CyclicBarrier;/** * CyclicBarrier测试 * * @author 奔波儿灞 * @version 1.0 */publicclassCyclicBarrierTest {privatefinalLoggerlogger =LoggerFactory.getLogger(ExchangerTest.class);@Testpublicvoidrun() {CyclicBarriercyclicBarrier =newCyclicBarrier(2);newThread(newRunnable() {@Overridepublicvoidrun() {logger.info("准备");try {cyclicBarrier.await(); }catch (InterruptedException |BrokenBarrierExceptione) {logger.warn("Thread cyclicBarrier await error",e); }logger.info("结束"); } }).start();logger.info("准备");try {cyclicBarrier.await(); }catch (InterruptedException |BrokenBarrierExceptione) {logger.warn("Thread cyclicBarrier await error",e); }logger.info("结束"); }}
输出:
11:11:18.690 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 准备11:11:18.690 [main] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 准备11:11:18.694 [Thread-0] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 结束11:11:18.694 [main] INFO tk.fishfish.easyjava.concurrent.ExchangerTest - 结束
CountDownLatch
允许一个或多个线程等待其他线程完成操作。
- 等待其他多线程执行完毕。(thread join)
packagetk.fishfish.easyjava.concurrent;importorg.junit.Test;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importjava.util.concurrent.CountDownLatch;/** * CountDownLatch使用 * * @author 奔波儿灞 * @version 1.0 */publicclassCountDownLatchTest {privatefinalLoggerlogger =LoggerFactory.getLogger(CountDownLatchTest.class);@Testpublicvoidrun() {CountDownLatchlatch =newCountDownLatch(2);// thread 1newThread(newRunnable() {@Overridepublicvoidrun() {logger.info("我开始执行...");latch.countDown();logger.info("我完成了!"); } }).start();// thread 1newThread(newRunnable() {@Overridepublicvoidrun() {logger.info("我开始执行...");latch.countDown();logger.info("我完成了!"); } }).start();logger.info("等待大家...");try {latch.await(); }catch (InterruptedExceptione) {logger.warn("Thread interrupted",e); }logger.info("大家都完成了!"); }}
输出:
11:18:00.870 [Thread-0] INFO tk.fishfish.easyjava.concurrent.CountDownLatchTest - 我开始执行...11:18:00.870 [main] INFO tk.fishfish.easyjava.concurrent.CountDownLatchTest - 等待大家...11:18:00.873 [Thread-0] INFO tk.fishfish.easyjava.concurrent.CountDownLatchTest - 我完成了!11:18:00.870 [Thread-1] INFO tk.fishfish.easyjava.concurrent.CountDownLatchTest - 我开始执行...11:18:00.873 [Thread-1] INFO tk.fishfish.easyjava.concurrent.CountDownLatchTest - 我完成了!11:18:00.873 [main] INFO tk.fishfish.easyjava.concurrent.CountDownLatchTest - 大家都完成了!
可以先看下官方文档,然后运行我之前写的例子,来进行入门。
当然,阅读其他关于kafka
的技术文章均可,但是建议过一遍官网。因为,技术文章可能过时啦!
这里要说一说spring-kafka
,因为封装的比较厉害,可能跟你实际使用起来有很大的差别。
在spring-boot
基础上添加依赖:
<dependency> <groupId>org.springframework.kafka</groupId> <artifactId>spring-kafka</artifactId> <version>2.2.3.RELEASE</version></dependency>
注意要检查下依赖的kakfa-clients
,是否与你服务端的匹配。
通过注解@EnableKafka
开启自动配置。
这里测试消费者,采用默认的配置:
spring:kafka:consumer:bootstrap-servers: -127.0.0.1:9092# 消费组group-id:myGroup# 消费者是否自动提交偏移量,默认为trueenable-auto-commit:false# 消费者在读取一个没有偏移量或者偏移量无效的情况下,从起始位置读取partition的记录,默认是latestauto-offset-reset:earliest# 单次调用poll方法能够返回的消息数量max-poll-records:50
然后写个测试用例试试:
packagetk.fishfish.easyjava.kafka;importorg.apache.kafka.clients.consumer.ConsumerRecord;importorg.junit.Test;importorg.junit.runner.RunWith;importorg.slf4j.Logger;importorg.slf4j.LoggerFactory;importorg.springframework.boot.test.context.SpringBootTest;importorg.springframework.kafka.annotation.KafkaListener;importorg.springframework.test.context.junit4.SpringRunner;/** * 消费者 * * @author 奔波儿灞 * @since 1.0 */@SpringBootTest@RunWith(SpringRunner.class)publicclassConsumerTest {privatestaticfinalLoggerLOG =LoggerFactory.getLogger(ConsumerTest.class);@KafkaListener(topics ="test")publicvoidonMessage(ConsumerRecord<String,String>record) {LOG.info("record: {}",record);Stringvalue =record.value();if (value.length() %2 ==0) {thrownewRuntimeException("模拟业务出错"); } }@Testpublicvoidrun() {try {// 阻塞5分钟,方便调试Thread.sleep(5 *60 *1000); }catch (InterruptedExceptione) {LOG.warn("sleep error",e); } }}
上面通过@KafkaListener
来监听topic
,处理消息。
为了模拟业务会出现一些异常,我特意在判断value长度为偶数的情况下抛出异常,看在默认配置的情况下,如果业务出错,是否仍会提交offsets
。
结果发现,仍提交了offsets。
下面是日志:
INFO 2019-05-17 10:25:48.557 [org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1] tk.fishfish.easyjava.kafka.ConsumerTest - record: ConsumerRecord(topic = test, partition = 0, offset = 261, CreateTime = 1558059947687, serialized key size = -1, serialized value size = 0, headers = RecordHeaders(headers = [], isReadOnly = false), key = null, value = )ERROR 2019-05-17 10:25:48.557 [org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1] o.s.kafka.listener.LoggingErrorHandler - Error while processing: ConsumerRecord(topic = test, partition = 0, offset = 261, CreateTime = 1558059947687, serialized key size = -1, serialized value size = 0, headers = RecordHeaders(headers = [], isReadOnly = false), key = null, value = )org.springframework.kafka.listener.ListenerExecutionFailedException: Listener method 'public void tk.fishfish.easyjava.kafka.ConsumerTest.onMessage(org.apache.kafka.clients.consumer.ConsumerRecord<java.lang.String, java.lang.String>)' threw exception; nested exception is java.lang.RuntimeException: 模拟业务出错 at org.springframework.kafka.listener.adapter.MessagingMessageListenerAdapter.invokeHandler(MessagingMessageListenerAdapter.java:302) at org.springframework.kafka.listener.adapter.RecordMessagingMessageListenerAdapter.onMessage(RecordMessagingMessageListenerAdapter.java:79) at org.springframework.kafka.listener.adapter.RecordMessagingMessageListenerAdapter.onMessage(RecordMessagingMessageListenerAdapter.java:50) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.doInvokeOnMessage(KafkaMessageListenerContainer.java:1220) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.invokeOnMessage(KafkaMessageListenerContainer.java:1213) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.doInvokeRecordListener(KafkaMessageListenerContainer.java:1174) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.doInvokeWithRecords(KafkaMessageListenerContainer.java:1155) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.invokeRecordListener(KafkaMessageListenerContainer.java:1096) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.invokeListener(KafkaMessageListenerContainer.java:924) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.pollAndInvoke(KafkaMessageListenerContainer.java:740) at org.springframework.kafka.listener.KafkaMessageListenerContainer$ListenerConsumer.run(KafkaMessageListenerContainer.java:689) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) at java.util.concurrent.FutureTask.run(FutureTask.java:266) at java.lang.Thread.run(Thread.java:748)Caused by: java.lang.RuntimeException: 模拟业务出错 at tk.fishfish.easyjava.kafka.ConsumerTest.onMessage(ConsumerTest.java:29) at sun.reflect.GeneratedMethodAccessor41.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:498) at org.springframework.messaging.handler.invocation.InvocableHandlerMethod.doInvoke(InvocableHandlerMethod.java:170) at org.springframework.messaging.handler.invocation.InvocableHandlerMethod.invoke(InvocableHandlerMethod.java:120) at org.springframework.kafka.listener.adapter.HandlerAdapter.invoke(HandlerAdapter.java:48) at org.springframework.kafka.listener.adapter.MessagingMessageListenerAdapter.invokeHandler(MessagingMessageListenerAdapter.java:283) ... 13 common frames omittedDEBUG 2019-05-17 10:25:48.557 [org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1] o.s.k.l.a.RecordMessagingMessageListenerAdapter - Processing [GenericMessage [payload=d, headers={kafka_offset=262, kafka_consumer=org.apache.kafka.clients.consumer.KafkaConsumer@f46d581, kafka_timestampType=CREATE_TIME, kafka_receivedMessageKey=null, kafka_receivedPartitionId=0, kafka_receivedTopic=test, kafka_receivedTimestamp=1558059947851}]]INFO 2019-05-17 10:25:48.557 [org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1] tk.fishfish.easyjava.kafka.ConsumerTest - record: ConsumerRecord(topic = test, partition = 0, offset = 262, CreateTime = 1558059947851, serialized key size = -1, serialized value size = 1, headers = RecordHeaders(headers = [], isReadOnly = false), key = null, value = d)DEBUG 2019-05-17 10:25:48.557 [org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1] o.s.k.l.KafkaMessageListenerContainer$ListenerConsumer - Commit list: {test-0=OffsetAndMetadata{offset=263, metadata=''}}DEBUG 2019-05-17 10:25:48.557 [org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1] o.s.k.l.KafkaMessageListenerContainer$ListenerConsumer - Committing: {test-0=OffsetAndMetadata{offset=263, metadata=''}}DEBUG 2019-05-17 10:25:48.560 [org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1] o.a.k.c.consumer.internals.ConsumerCoordinator - [Consumer clientId=consumer-2, groupId=myGroup] Committed offset 263 for partition test-0
从日志可以看到offset = 261
的记录处理失败了,但最后仍提交了Committed offset 263 for partition test-0
。
总结:如果你采取我这样的配置,当处理record
出错的时候,仍会提交偏移量。那么我们就需要业务处理失败的情况了。比如try...catch
之后保存错误的record
,然后定时重试。
那么,能不能在出错的情况下不提交咧?
通过查看文档发现,发现可以使用Acknowledgment
去确认该条record
是否提交。
修改下配置,配置spring.kafka.listener.*
:
spring:kafka:consumer:bootstrap-servers: -127.0.0.1:9092group-id:myGroup# 消费者是否自动提交偏移量,默认为trueenable-auto-commit:false# 消费者在读取一个没有偏移量或者偏移量无效的情况下,从起始位置读取partition的记录,默认是latestauto-offset-reset:earliest# 单次调用poll方法能够返回的消息数量max-poll-records:50listener:# Listener AckModeack-mode:MANUAL_IMMEDIATE# 并发消费者concurrency:1
然后代码有些调整(就不贴全了),使用Acknowledgment
:
@KafkaListener(topics ="test")publicvoidonMessage(ConsumerRecord<String,String>record,Acknowledgmentack) {LOG.info("record: {}",record);Stringvalue =record.value();if (value.length() %2 ==0) {thrownewRuntimeException("模拟业务出错"); }// 业务处理成功确认ack.acknowledge();}
如果在业务出错的情况下,不会提交offsets
,然而真的是这样的吗?
测试发现,在业务出错的情况下,确实不会提交offsets
,但是只要后面的记录处理成功,就会提交offsets
,这样前面的失败的数据还是需要自己去手动处理。要么重新获取该offset
的数据,要么记录错误record
,业务重试。
待续
待续
JVM诊断工具。
计算方法调用时常,可用来分析性能较慢的方法调用。
参看方法参数,返回值。可通过 -x 参数查看类更深层次信息。
查看线程信息。
查看阻塞的线程
thread -b
查看最繁忙的线程(top-10)
thread -n 10
查看加载的类信息。
查看类方法。
反编译。
以上内容,如果有错误,欢迎批评指正。如果在工作或学习上帮到了你,给个star哟!