akalash commented on a change in pull request #18332:
URL: https://github.com/apache/flink/pull/18332#discussion_r783137704
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/minicluster/MiniCluster.java
##########
@@ -1175,6 +1185,31 @@ private void terminateMiniClusterServices() throws
Exception {
}
}
+ /**
+ * Prevent multiple submission of the same JobGraph that has been mutated
in between
Review comment:
I agree that the contract looks strange. Why can not we indeed clone the
JobGraph in submitJob?(it is better not to clone but use builder and so but as
I understand it is not easy to do so now)
##########
File path:
flink-tests/src/test/java/org/apache/flink/test/checkpointing/SavepointITCase.java
##########
@@ -431,36 +430,48 @@ public void invoke(Long value) {
.build());
cluster.before();
try {
- final JobID jobID1 = new JobID();
- jobGraph.setJobID(jobID1);
- cluster.getClusterClient().submitJob(jobGraph).get();
- CommonTestUtils.waitForAllTaskRunning(cluster.getMiniCluster(),
jobID1, false);
+ final JobID firstJobId = new JobID();
+ final JobGraph firstJobGraph = InstantiationUtil.clone(jobGraph);
+ firstJobGraph.setJobID(firstJobId);
+ cluster.getClusterClient().submitJob(firstJobGraph).get();
+ CommonTestUtils.waitForAllTaskRunning(cluster.getMiniCluster(),
firstJobId, false);
// wait for some records to be processed before taking the
checkpoint
counter.get().await();
- final String firstCheckpoint =
cluster.getMiniCluster().triggerCheckpoint(jobID1).get();
-
- cluster.getClusterClient().cancel(jobID1).get();
- jobGraph.setSavepointRestoreSettings(
+ final String firstCheckpoint =
+
cluster.getMiniCluster().triggerCheckpoint(firstJobId).get();
+ cluster.getClusterClient().cancel(firstJobId).get();
+ CommonTestUtils.waitForJobStatus(
Review comment:
I don't fully understand why we need it here. if we clone JobGraph then
everything works perfectly fine even without this waiting. It seems for me that
the awaiting on `cancel's future` is enough to be sure that the next submit
with the new JobGraph will be successful. Did I miss some rare race condition?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
