diff --git a/temporal-sdk/src/test/java/io/temporal/client/functional/MetricsTest.java b/temporal-sdk/src/test/java/io/temporal/client/functional/MetricsTest.java
index 98fddf17c..ce353e940 100644
--- a/temporal-sdk/src/test/java/io/temporal/client/functional/MetricsTest.java
+++ b/temporal-sdk/src/test/java/io/temporal/client/functional/MetricsTest.java
@@ -63,6 +63,7 @@ public class MetricsTest {
 
   @Before
   public void setUp() {
+    registry.clear();
     tagsNamespaceQueue =
         replaceTags(TAGS_NAMESPACE, MetricsTag.TASK_QUEUE, testWorkflowRule.getTaskQueue());
   }
diff --git a/temporal-test-server/src/main/java/io/temporal/internal/testservice/TestWorkflowMutableStateImpl.java b/temporal-test-server/src/main/java/io/temporal/internal/testservice/TestWorkflowMutableStateImpl.java
index 61d29ece6..c667c47a2 100644
--- a/temporal-test-server/src/main/java/io/temporal/internal/testservice/TestWorkflowMutableStateImpl.java
+++ b/temporal-test-server/src/main/java/io/temporal/internal/testservice/TestWorkflowMutableStateImpl.java
@@ -488,14 +488,16 @@ public void completeWorkflowTask(
 
           if (unhandledCommand(request) || unhandledMessages(request)) {
             // Fail the workflow task if there are new events or messages and a command tries to
-            // complete the
-            // workflow
+            // complete the workflow. Record the failure in history, then throw an error to the
+            // caller (matching real server behavior).
             failWorkflowTaskWithAReason(
                 WorkflowTaskFailedCause.WORKFLOW_TASK_FAILED_CAUSE_UNHANDLED_COMMAND,
                 null,
                 ctx,
                 request,
                 false);
+            ctx.setExceptionIfEmpty(
+                Status.INVALID_ARGUMENT.withDescription("UnhandledCommand").asRuntimeException());
             return;
           }