Github user yuqi1129 commented on a diff in the pull request:
https://github.com/apache/flink/pull/5846#discussion_r191143746
--- Diff:
flink-clients/src/main/java/org/apache/flink/client/program/ClusterClient.java
---
@@ -896,6 +898,10 @@ public static JobGraph getJobGraph(Configuration
flinkConfig, FlinkPlan optPlan,
job = gen.compileJobGraph((OptimizedPlan) optPlan);
}
+ // if we disable checkpoint and do not set restart strategy,
Restart strategy will be set as in flink-conf.yaml
+ // in flip6, jobmaster do not set this conf, so we have set
this conf here.
--- End diff --
before flip6, cluster site will set restart strategy:
```
val restartStrategy =
Option(jobGraph.getSerializedExecutionConfig()
.deserializeValue(userCodeLoader)
.getRestartStrategy())
.map(RestartStrategyFactory.createRestartStrategy)
.filter(p => p != null) match {
case Some(strategy) => strategy
case None => restartStrategyFactory.createRestartStrategy()
}
```
in Jobmanger.java
in flip6, cluster will directly use jobgraph deserialized from client size
---
