Bug fix: pass along system descriptor when creating a failure index

Author: gmarouli

server/src/main/java/org/elasticsearch/action/admin/indices/rollover/MetadataRolloverService.java

@@ -360,6 +360,7 @@ private RolloverResult rolloverDataStream(
                 projectState.cluster(),
                 now.toEpochMilli(),
                 dataStreamName,
+                systemDataStreamDescriptor,
                 templateV2,
                 newWriteIndexName,
                 (builder, indexMetadata) -> builder.put(dataStream.rolloverFailureStore(indexMetadata.getIndex(), newGeneration))

server/src/main/java/org/elasticsearch/cluster/metadata/MetadataCreateDataStreamService.java

@@ -290,6 +290,7 @@ static ClusterState createDataStream(
                 currentState,
                 request.startTime(),
                 dataStreamName,
+                systemDataStreamDescriptor,
                 template,
                 failureStoreIndexName,
                 null
@@ -419,6 +420,7 @@ public static ClusterState createFailureStoreIndex(
         ClusterState currentState,
         long nameResolvedInstant,
         String dataStreamName,
+        SystemDataStreamDescriptor systemDataStreamDescriptor,
         ComposableIndexTemplate template,
         String failureStoreIndexName,
         @Nullable BiConsumer<ProjectMetadata.Builder, IndexMetadata> metadataTransformer
@@ -439,7 +441,8 @@ public static ClusterState createFailureStoreIndex(
             .performReroute(false)
             .setMatchingTemplate(template)
             .settings(indexSettings)
-            .isFailureIndex(true);
+            .isFailureIndex(true)
+            .systemDataStreamDescriptor(systemDataStreamDescriptor);
 
         try {
             currentState = metadataCreateIndexService.applyCreateIndexRequest(