一、概述
当任务失败时，Flink需要重新启动失败的任务和其他受影响的任务，将作业恢复到正常状态；重新启动策略和故障转移策略用于控制任务的重新启动。重新启动策略决定是否以及何时可以重新启动失败/受影响的任务。故障转移策略决定应该重新启动哪些任务以恢复作业。
二、Restart Strategies 重启策略
在没有定义特定作业的重启策略时，总是使用默认的重启策略。如果提交的作业带有重启策略，该策略将覆盖集群的默认设置。
默认的重启策略是通过Flink的配置文件flink-conf.yml设置的，也可以通过代码设置覆盖默认设置，重启策略主要分了以下几种类型
配置参数restart-strategy定义了采取哪种策略。如果没有启用Checkpoint，则使用“no restart”策略；如果启用了 checkpointing，但没有配置重启策略，则使用固定间隔 (fixed-delay) 策略，其中 Integer.MAX_VALUE 参数是尝试重启次数，
restart-strategy重启策略主要分了以下几种类型：
1.none, off, disable
不设置重启策略，如果没有启用Checkpoint，则使用“no restart”策略；
a.通过配置文件flink-conf.yaml设置：
restart-strategy: none b.通过代码设置：
ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();env.setRestartStrategy(RestartStrategies.noRestart()); 2.fixeddelay, fixed-delay
固定时间间隔，可设置重启次数和重试时间间隔，如果启用了 checkpointing，但没有配置重启策略，则使用固定间隔 (fixed-delay) 策略；
restart-strategy: fixed-delay restart-strategy.fixed-delay.attempts: 3restart-strategy.fixed-delay.delay: 10 s ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();env.setRestartStrategy(RestartStrategies.fixedDelayRestart( 3, // 失败重启次数 Time.of(10, TimeUnit.SECONDS) //每次重启时间间隔)); 3.failurerate, failure-rate:
失败率重启策略在Job失败后会重启，但是超过失败率后，Job会最终被认定失败。在两个连续的重启尝试之间，重启策略会等待一个固定的时间.如下配置是5分钟内若失败了3次则认为该job失败，重试间隔为10s。
restart-strategy: failure-raterestart-strategy.failure-rate.max-failures-per-interval: 3restart-strategy.failure-rate.failure-rate-interval: 5 minrestart-strategy.failure-rate.delay: 10 s ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();env.setRestartStrategy(RestartStrategies.failureRateRestart( 3, // max failures per interval Time.of(5, TimeUnit.MINUTES), //time interval for measuring failure rate Time.of(10, TimeUnit.SECONDS) // delay)); 三、重启策略代码演示
这里还是从Socket端接收数据，然后进行单词统计，主函数代码基本不变，收到"error"开头的数据，直接抛出异常，各种情况修改Checkpoint的相关设置进行功能演示：
1.第一种情况，不设置Checkpoint，默认是无重启策略，代码如下：
package com.hadoop.ljs.flink110.checkpoint;import org.apache.flink.api.common.functions.FilterFunction;import org.apache.flink.api.common.functions.FlatMapFunction;import org.apache.flink.api.java.tuple.Tuple2;import org.apache.flink.streaming.api.datastream.DataStream;import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;import org.apache.flink.util.Collector;/** * @author: Created By lujisen * @company ChinaUnicom Software JiNan * @date: 2020-04-12 09:26 * @version: v1.0 * @description: com.hadoop.ljs.flink110.checkpoint */public class CheckpointTest { public static void main(String[] args) throws Exception { /*初始化Flink流处理环境*/ StreamExecutionEnvironment senv = StreamExecutionEnvironment.getExecutionEnvironment(); /*从socket接收数据*/ DataStream<String> source = senv.socketTextStream("localhost", 9000); DataStream<String> filterData=source.filter(new FilterFunction<String>() { @Override public boolean filter(String value) throws Exception { if(null==value||"".equals(value)){ return false; } return true; } }); DataStream<Tuple2<String, Integer>> wordOne = filterData.flatMap(new FlatMapFunction<String, Tuple2<String, Integer>>() { @Override public void flatMap(String value, Collector<Tuple2<String, Integer>> out) throws Exception { /*收到error开头的数据 手动抛出异常 进行测试*/ if(value.startsWith("error")){ throw new Exception("收到error，报错！！！"); } String[] lines = value.split(","); for(int i=0;i<lines.length;i++){ Tuple2<String,Integer> wordTmp=new Tuple2<>(lines[i],1); out.collect(wordTmp); } } }); DataStream<Tuple2<String, Integer>> wordCount = wordOne.keyBy(0).sum(1); wordCount.print(); senv.execute(); }} 这里收到error信息，抛出异常,无重启策略情况下，整个job直接挂掉。
2.第二种情况，启用Checkpoint，采用固定时间间隔重启策略，代码如下：
package com.hadoop.ljs.flink110.checkpoint;import akka.remote.WireFormats;import org.apache.flink.api.common.functions.FilterFunction;import org.apache.flink.api.common.functions.FlatMapFunction;import org.apache.flink.api.common.restartstrategy.RestartStrategies;import org.apache.flink.api.common.time.Time;import org.apache.flink.api.java.tuple.Tuple2;import org.apache.flink.streaming.api.datastream.DataStream;import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;import org.apache.flink.util.Collector;/** * @author: Created By lujisen * @company ChinaUnicom Software JiNan * @date: 2020-04-12 09:26 * @version: v1.0 * @description: com.hadoop.ljs.flink110.checkpoint */public class CheckpointTest { public static void main(String[] args) throws Exception { /*初始化Flink流处理环境*/ StreamExecutionEnvironment senv = StreamExecutionEnvironment.getExecutionEnvironment(); /*每隔10S做一次checkpoint*/ senv.enableCheckpointing(10000); /*采用固定时间间隔重启策略，设置重启次数和重启时间间隔*/ senv.setRestartStrategy(RestartStrategies.fixedDelayRestart( 3,// 尝试重启的次数 Time.seconds(2) //重启时间间隔 )); /*从socket接收数据*/ DataStream<String> source = senv.socketTextStream("localhost", 9000); DataStream<String> filterData=source.filter(new FilterFunction<String>() { @Override public boolean filter(String value) throws Exception { if(null==value||"".equals(value)){ return false; } return true; } }); DataStream<Tuple2<String, Integer>> wordOne = filterData.flatMap(new FlatMapFunction<String, Tuple2<String, Integer>>() { @Override public void flatMap(String value, Collector<Tuple2<String, Integer>> out) throws Exception { /*收到error开头的数据 手动抛出异常 进行测试*/ if(value.startsWith("error")){ throw new Exception("收到error，报错！！！"); } String[] lines = value.split(","); for(int i=0;i<lines.length;i++){ Tuple2<String,Integer> wordTmp=new Tuple2<>(lines[i],1); out.collect(wordTmp); } } }); DataStream<Tuple2<String, Integer>> wordCount = wordOne.keyBy(0).sum(1); wordCount.print(); senv.execute(); }} 设定了重启次数3次，重启时间间隔2秒，这里我发送前3次error，主程序自动重启不会挂掉，当我重新发送第4次error的时候主程序达到重启上限，直接挂掉了，如图所示：
3.第三种情况，启用Checkpoint，设置失败率重启策略，并指定2分钟时间内，最大失败次数为3次，重启时间间隔10秒，代码如下：
package com.hadoop.ljs.flink110.checkpoint;import akka.remote.WireFormats;import org.apache.flink.api.common.functions.FilterFunction;import org.apache.flink.api.common.functions.FlatMapFunction;import org.apache.flink.api.common.restartstrategy.RestartStrategies;import org.apache.flink.api.common.time.Time;import org.apache.flink.api.java.ExecutionEnvironment;import org.apache.flink.api.java.tuple.Tuple2;import org.apache.flink.streaming.api.datastream.DataStream;import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;import org.apache.flink.util.Collector;/** * @author: Created By lujisen * @company ChinaUnicom Software JiNan * @date: 2020-04-12 09:26 * @version: v1.0 * @description: com.hadoop.ljs.flink110.checkpoint */public class CheckpointTest { public static void main(String[] args) throws Exception { /*初始化Flink流处理环境*/ StreamExecutionEnvironment senv = StreamExecutionEnvironment.getExecutionEnvironment(); /*每隔10S做一次checkpoint*/ senv.enableCheckpointing(10000); /* 设置失败率重启策略，下面设置2分钟内 最大失败次数3次 重试时间间隔10秒*/ senv.setRestartStrategy(RestartStrategies.failureRateRestart( 3, // Time.minutes(2), //time interval for measuring failure rate Time.seconds(10) // delay )); /*从socket接收数据*/ DataStream<String> source = senv.socketTextStream("localhost", 9000); DataStream<String> filterData=source.filter(new FilterFunction<String>() { @Override public boolean filter(String value) throws Exception { if(null==value||"".equals(value)){ return false; } return true; } }); DataStream<Tuple2<String, Integer>> wordOne = filterData.flatMap(new FlatMapFunction<String, Tuple2<String, Integer>>() { @Override public void flatMap(String value, Collector<Tuple2<String, Integer>> out) throws Exception { /*收到error开头的数据 手动抛出异常 进行测试*/ if(value.startsWith("error")){ throw new Exception("收到error，报错！！！"); } String[] lines = value.split(","); for(int i=0;i<lines.length;i++){ Tuple2<String,Integer> wordTmp=new Tuple2<>(lines[i],1); out.collect(wordTmp); } } }); DataStream<Tuple2<String, Integer>> wordCount = wordOne.keyBy(0).sum(1); wordCount.print(); senv.execute(); }} 这里使用失败率重启策略，2分钟内，收到第三次error时，主程序达到失败次数，直接挂掉。
![Java小案例——随机数猜测随机数猜测[图]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)