Savepoint 在 Flink On Yarn 下的使用

发表于 | 更新于 | 分类于 | 评论数：

配置文件

在 flink-conf.yaml 中配置 savepoint 的存储位置

1
state.savepoints.dir: hdfs:///flink/flink-checkpoints

列出当前job

1
2
3
flink list -yid [for yarn, applicationId] -z [zookeepernamesapce, for high availablity mode]

flink list -yid application_1541130324309_128080 -z cluster_by_job

停止job，并存储状态

1
2
3
flink cancel -s JobId -yid [for yarn, applicationId] -z [zookeepernamespace, for ha] 

flink cancel -s  cababd2aadf69a263eda49df0f1efa82 -yid application_1541130324309_126267 -z cluster_by_job

从指定 savepoint 启动 job

1
flink run -s  hdfs:///flink/flink-checkpoints/savepoint-cababd-b92695040cfe  [args]

强力建议