Flink状态管理与Checkpoint实战——模拟电商订单计算过程中宕机的场景，探索宕机恢复时如何精准继续计算订单（上）

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
//设置checkpoint的周期, 每隔1000 ms进行启动一个检查点
env.getCheckpointConfig().setCheckpointInterval(1000);
// 设置状态级别模式为exactly-once
env.getCheckpointConfig().setCheckpointingMode(CheckpointingMode.EXACTLY_ONCE);
//超时时间，可能是保存太耗费时间或者是状态后端的问题，任务同步执行不能一直阻塞
env.getCheckpointConfig().setCheckpointTimeout(60000L);  
// 设置取消和故障时是否保留Checkpoint数据，这个设置较为重要，没有正确的选择好可能会导致检查点数据失效
//有两个参数可以设置
//ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION： 取消作业时保留检查点。必须在取消后手动清理检查点状态。
//ExternalizedCheckpointCleanup.DELETE_ON_CANCELLATION： 取消作业时删除检查点。只有在作业失败时，检查点状态才可用。
env.getCheckpointConfig().enableExternalizedCheckpoints(CheckpointConfig.ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION);

public class VideoOrderSourceV2 extends RichParallelSourceFunction<VideoOrder> {
    private volatile Boolean flag = true;
    private Random random = new Random();
    private static List<VideoOrder> list = new ArrayList<>();
    static {
        list.add(new VideoOrder("","java",10,0,null));
        list.add(new VideoOrder("","spring boot",15,0,null));
    }
    /**
     * run 方法调用前 用于初始化连接
     * @param parameters
     * @throws Exception
     */
    @Override
    public void open(Configuration parameters) throws Exception {
        System.out.println("-----open-----");
    }
    /**
     * 用于清理之前
     * @throws Exception
     */
    @Override
    public void close() throws Exception {
        System.out.println("-----close-----");
    }
    /**
     * 产生数据的逻辑
     * @param ctx
     * @throws Exception
     */
    @Override
    public void run(SourceContext<VideoOrder> ctx) throws Exception {
        while (flag){
            Thread.sleep(1000);
            String id = UUID.randomUUID().toString().substring(30);
            int userId = random.nextInt(10);
            int videoNum = random.nextInt(list.size());
            VideoOrder videoOrder = list.get(videoNum);
            videoOrder.setUserId(userId);
            videoOrder.setCreateTime(new Date());
            videoOrder.setTradeNo(id);
            System.out.println("产生:"+videoOrder.getTitle()+"，价格:"+videoOrder.getMoney()+", 时间:"+ TimeUtil.format(videoOrder.getCreateTime()));
            ctx.collect(videoOrder);
        }
    }
    /**
     * 控制任务取消
     */
    @Override
    public void cancel() {
        flag = false;
    }
}

public class FlinkKeyByReduceApp {
    /**
     * source
     * transformation
     * sink
     *
     * @param args
     */
    public static void main(String[] args) throws Exception {
        //构建执行任务环境以及任务的启动的入口, 存储全局相关的参数
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setParallelism(1);
        env.getCheckpointConfig().setCheckpointingMode(CheckpointingMode.EXACTLY_ONCE);
        env.enableCheckpointing(5000);
 env.getCheckpointConfig().enableExternalizedCheckpoints(CheckpointConfig.ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION);
        //这是我本机的ip地址       
        env.getCheckpointConfig().setCheckpointStorage(new                                           FileSystemCheckpointStorage("hdfs://192.168.192.100:8020/checkpoint"));
       DataStreamSource<VideoOrder> ds = env.addSource(new VideoOrderSourceV2());
        KeyedStream<VideoOrder, String> videoOrderStringKeyedStream = ds.keyBy(new KeySelector<VideoOrder, String>() {
            @Override
            public String getKey(VideoOrder value) throws Exception {
                return value.getTitle();
            }
        });
        SingleOutputStreamOperator<VideoOrder> reduce = videoOrderStringKeyedStream.reduce(new ReduceFunction<VideoOrder>() {
            @Override
            public VideoOrder reduce(VideoOrder value1, VideoOrder value2) throws Exception {
                VideoOrder videoOrder = new VideoOrder();
                videoOrder.setTitle(value1.getTitle());
                videoOrder.setMoney(value1.getMoney() + value2.getMoney());
                return videoOrder;
            }
        });
        reduce.print();
        env.execute("job");
    }
}

-s : 指定检查点的元数据的位置，这个位置记录着宕机前程序的计算状态
./bin/flink run -s /checkpoint/id号/chk-23/_metadata -c net.xxx.xxx.FlinkKeyByReduceApp -p 3 /root/xdclass-flink.jar