Github user zentol commented on a diff in the pull request:
https://github.com/apache/flink/pull/5542#discussion_r169714628
--- Diff:
flink-tests/src/test/java/org/apache/flink/test/streaming/runtime/BigUserProgramJobSubmitITCase.java
---
@@ -43,40 +40,27 @@
/**
* Integration test that verifies that a user program with a big(ger)
payload is successfully
* submitted and run.
+ *
+ * <p>This test fails for non-flip6 cluster due to job submission payload
being too large, see [FLINK-7285].
*/
-@Ignore("Fails on job submission payload being too large - [FLINK-7285]")
+@Category(Flip6.class)
public class BigUserProgramJobSubmitITCase extends TestLogger {
//
------------------------------------------------------------------------
// The mini cluster that is shared across tests
//
------------------------------------------------------------------------
- private static final int DEFAULT_PARALLELISM = 1;
-
- private static LocalFlinkMiniCluster cluster;
-
- private static final Logger LOG =
LoggerFactory.getLogger(BigUserProgramJobSubmitITCase.class);
+ @ClassRule
+ public static final MiniClusterResource MINI_CLUSTER_RESOURCE = new
MiniClusterResource(
+ new MiniClusterResource.MiniClusterResourceConfiguration(
+ new Configuration(),
+ 1,
+ 1));
//
------------------------------------------------------------------------
// Cluster setup & teardown
//
------------------------------------------------------------------------
- @BeforeClass
- public static void setup() throws Exception {
- // make sure we do not use a singleActorSystem for the tests
- // (therefore, we cannot simply inherit from
StreamingMultipleProgramsTestBase)
--- End diff --
maybe it does succeed exactly because of that. If everything runs in the
same actorSystem the messagesa ren't subject to the `akka.framesize` limit.
---
