Github user dawidwys commented on a diff in the pull request:
https://github.com/apache/flink/pull/5846#discussion_r191133623
--- Diff:
flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/Dispatcher.java
---
@@ -255,6 +264,35 @@ public void start() throws Exception {
}
}
+ /**
+ *
+ * @param job job graph
+ * @param configuration configuration in flink-conf.yaml
+ * @param isClusterPoint whether this is client side or cluster site
+ */
+ public static void setJobgraphRestartStrategy(JobGraph job,
Configuration configuration, boolean isClusterPoint) {
+
+ try {
+ ExecutionConfig config =
job.getSerializedExecutionConfig().deserializeValue(
+ Dispatcher.class.getClassLoader());
+
+ if (config.getRestartStrategy() == null) {
+ RestartStrategies.RestartStrategyConfiguration
restartStrategyConfiguration =
+
RestartStrategyFactory.createRestartStrategyConfiguration(configuration);
+
+ //FixedDelay will be as default
+ if (restartStrategyConfiguration == null &&
isClusterPoint) {
--- End diff --
I think in case the checkpointing is disabled the default strategy should
be `noRestart`.
---
