flink的savepoint实验-java

前言和一些准备工作

首先需要明确的是，savepoint和checkpoint不是同一类，

所以在进行savepoint实验时，代码里不需要添加checkpoint enable选项等．

$FLINK_HOME/conf/flink-conf.yaml中添加

state.savepoints.dir: hdfs://Desktop:9000/flink-savepoints

实验代码采用[2]

------------------------------------------------------------------------------------------------------------------------

实验步骤

首先注意，这个实验在真实集群的standalone中进行，而不是local单节伪分布式集群．

步骤	步骤内容
①	启动Zookeeper,Hadoop,Flink(HA)三种集群(启动顺序不要错)
②	mvn package cd target 扫描二维码关注公众号，回复： 11845388 查看本文章 flink run -c WordCount /home/appleyuchi/桌面/Flink_Code/flink_state/savepoint/target/bbb-1.0-SNAPSHOT.jar 得到: Job has been submitted with JobID a5c47eb5cd62ef61f0e0ae66f17aab58
③	flink savepoint a5c47eb5cd62ef61f0e0ae66f17aab58 Savepoint completed. Path: hdfs://Desktop:9000/flink-savepoints/savepoint-a5c47e-bd57d19f884c You can resume your program from this savepoint with the run command.
④	flink cancel a5c47eb5cd62ef61f0e0ae66f17aab58 Cancelling job a5c47eb5cd62ef61f0e0ae66f17aab58. Cancelled job a5c47eb5cd62ef61f0e0ae66f17aab58. WEB UI显示
⑤	flink run -s hdfs://Desktop:9000/flink-savepoints/savepoint-a5c47e-bd57d19f884c  -c WordCount -d /home/appleyuchi/桌面/Flink_Code/flink_state/savepoint/target/bbb-1.0-SNAPSHOT.jar Job has been submitted with JobID 00b79db0e9123ebda8dcbb22a8d1e484 通过上述命令恢复以后，会在web ui中看到restore信息被更新 同时，之前Task Managers中的结果依然存在 我们继续在nc -lk 9999中输入heiheihei，能不能在上述同一个taskmanager中继续输出呢？答案是不能， restore以后的程序在输出结果时，跑到另外一个taskmanager中去了,所以这点有些不完美

上述流程主要参考了[1]

完整代码

Reference

[1]Flink SavePoint

[2]intellij运行flink的wordcount实验-Java版本