Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/5771#discussion_r177375838
--- Diff:
flink-tests/src/test/java/org/apache/flink/test/accumulators/AccumulatorLiveITCase.java
---
@@ -167,6 +172,9 @@ private static void submitJobAndVerifyResults(JobGraph
jobGraph) throws Exceptio
NotifyingMapper.shutdownLatch.trigger();
} finally {
NotifyingMapper.shutdownLatch.trigger();
+
+ // wait for the completion of the job
+
clusterClient.requestJobResult(jobGraph.getJobID()).get();
--- End diff --
It is true that this introduces a new pattern, but I think it reflects more
how the future `ClusterClient` abstraction will work. The `RestClusterClient`
already has these methods and uses them to implement the
`ClusterClient#submitJob` call.
I think it is not a big problem that some of the tests use polling and this
one waits on the future. They can be refactored once we introduce the new
`ClusterClient` interface. Moreover, this tests runs explicitly only with
Flip-6.
---
