MINOR: KAFKA-2371 follow-up, DistributedHerder should wakeup WorkerGroupMember after assignment to ensure work is started immediately

Author: Ewen Cheslack-Postava <me@ewencp.org>

Reviewers: Gwen Shapira

Closes #360 from ewencp/minor-kafka-2371-follow-up-wakeup-after-rebalance

copycat/runtime/src/main/java/org/apache/kafka/copycat/runtime/distributed/DistributedHerder.java

@@ -614,6 +614,10 @@ public class DistributedHerder implements Herder, Runnable {
                 log.info("Joined group and got assignment: {}", assignment);
                 DistributedHerder.this.assignment = assignment;
                 rebalanceResolved = false;
+                // We *must* interrupt any poll() call since this could occur when the poll starts, and we might then
+                // sleep in the poll() for a long time. Forcing a wakeup ensures we'll get to process this event in the
+                // main thread.
+                member.wakeup();
             }
 
             @Override

copycat/runtime/src/test/java/org/apache/kafka/copycat/runtime/distributed/DistributedHerderTest.java

@@ -371,6 +371,8 @@ public class DistributedHerderTest {
                 return null;
             }
         });
+        member.wakeup();
+        PowerMock.expectLastCall();
     }
 
     private void expectPostRebalanceCatchup(final ClusterConfigState readToEndSnapshot) {