resource/aws_elasticache_replication_group: Allow changing replica count in cluster-mode

.changelog/17301.txt

@@ -0,0 +1,3 @@
+```release-note:enhancement
+resource/aws_elasticache_replication_group: Allow changing `cluster_mode.replica_count` without re-creation
+```

aws/resource_aws_elasticache_replication_group.go

@@ -79,12 +79,8 @@ func resourceAwsElasticacheReplicationGroup() *schema.Resource {
 				Computed: true,
 			},
 			"cluster_mode": {
-				Type:     schema.TypeList,
-				Optional: true,
-				// We allow Computed: true here since using number_cache_clusters
-				// and a cluster mode enabled parameter_group_name will create
-				// a single shard replication group with number_cache_clusters - 1
-				// read replicas. Otherwise, the resource is marked ForceNew.
+				Type:         schema.TypeList,
+				Optional:     true,
 				Computed:     true,
 				MaxItems:     1,
 				ExactlyOneOf: []string{"cluster_mode", "number_cache_clusters"},
@@ -93,7 +89,6 @@ func resourceAwsElasticacheReplicationGroup() *schema.Resource {
 						"replicas_per_node_group": {
 							Type:     schema.TypeInt,
 							Required: true,
-							ForceNew: true,
 						},
 						"num_node_groups": {
 							Type:     schema.TypeInt,
@@ -314,7 +309,9 @@ func resourceAwsElasticacheReplicationGroup() *schema.Resource {
 				return nil
 			},
 			customdiff.ComputedIf("member_clusters", func(ctx context.Context, diff *schema.ResourceDiff, meta interface{}) bool {
-				return diff.HasChange("number_cache_clusters") || diff.HasChange("cluster_mode.0.num_node_groups")
+				return diff.HasChange("number_cache_clusters") ||
+					diff.HasChange("cluster_mode.0.num_node_groups") ||
+					diff.HasChange("cluster_mode.0.replicas_per_node_group")
 			}),
 		),
 	}
@@ -570,7 +567,7 @@ func resourceAwsElasticacheReplicationGroupRead(d *schema.ResourceData, meta int
 func resourceAwsElasticacheReplicationGroupUpdate(d *schema.ResourceData, meta interface{}) error {
 	conn := meta.(*AWSClient).elasticacheconn
 
-	if d.HasChange("cluster_mode.0.num_node_groups") {
+	if d.HasChanges("cluster_mode.0.num_node_groups", "cluster_mode.0.replicas_per_node_group") {
 		err := elasticacheReplicationGroupModifyShardConfiguration(conn, d)
 		if err != nil {
 			return fmt.Errorf("error modifying ElastiCache Replication Group (%s) shard configuration: %w", d.Id(), err)
@@ -761,6 +758,24 @@ func flattenElasticacheNodeGroupsToClusterMode(nodeGroups []*elasticache.NodeGro
 }
 
 func elasticacheReplicationGroupModifyShardConfiguration(conn *elasticache.ElastiCache, d *schema.ResourceData) error {
+	if d.HasChange("cluster_mode.0.num_node_groups") {
+		err := elasticacheReplicationGroupModifyShardConfigurationNumNodeGroups(conn, d)
+		if err != nil {
+			return err
+		}
+	}
+
+	if d.HasChange("cluster_mode.0.replicas_per_node_group") {
+		err := elasticacheReplicationGroupModifyShardConfigurationReplicasPerNodeGroup(conn, d)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+func elasticacheReplicationGroupModifyShardConfigurationNumNodeGroups(conn *elasticache.ElastiCache, d *schema.ResourceData) error {
 	o, n := d.GetChange("cluster_mode.0.num_node_groups")
 	oldNumNodeGroups := o.(int)
 	newNumNodeGroups := n.(int)
@@ -796,6 +811,44 @@ func elasticacheReplicationGroupModifyShardConfiguration(conn *elasticache.Elast
 	return nil
 }
 
+func elasticacheReplicationGroupModifyShardConfigurationReplicasPerNodeGroup(conn *elasticache.ElastiCache, d *schema.ResourceData) error {
+	o, n := d.GetChange("cluster_mode.0.replicas_per_node_group")
+	oldReplicas := o.(int)
+	newReplicas := n.(int)
+
+	if newReplicas > oldReplicas {
+		input := &elasticache.IncreaseReplicaCountInput{
+			ApplyImmediately:   aws.Bool(true),
+			NewReplicaCount:    aws.Int64(int64(newReplicas)),
+			ReplicationGroupId: aws.String(d.Id()),
+		}
+		_, err := conn.IncreaseReplicaCount(input)
+		if err != nil {
+			return fmt.Errorf("error adding ElastiCache Replication Group (%s) replicas: %w", d.Id(), err)
+		}
+		_, err = waiter.ReplicationGroupAvailable(conn, d.Id(), d.Timeout(schema.TimeoutUpdate))
+		if err != nil {
+			return fmt.Errorf("error waiting for ElastiCache Replication Group (%s) replica addition: %w", d.Id(), err)
+		}
+	} else {
+		input := &elasticache.DecreaseReplicaCountInput{
+			ApplyImmediately:   aws.Bool(true),
+			NewReplicaCount:    aws.Int64(int64(newReplicas)),
+			ReplicationGroupId: aws.String(d.Id()),
+		}
+		_, err := conn.DecreaseReplicaCount(input)
+		if err != nil {
+			return fmt.Errorf("error removing ElastiCache Replication Group (%s) replicas: %w", d.Id(), err)
+		}
+		_, err = waiter.ReplicationGroupAvailable(conn, d.Id(), d.Timeout(schema.TimeoutUpdate))
+		if err != nil {
+			return fmt.Errorf("error waiting for ElastiCache Replication Group (%s) replica removal: %w", d.Id(), err)
+		}
+	}
+
+	return nil
+}
+
 func elasticacheReplicationGroupModifyNumCacheClusters(conn *elasticache.ElastiCache, d *schema.ResourceData) error {
 	o, n := d.GetChange("number_cache_clusters")
 	oldNumberCacheClusters := o.(int)

aws/resource_aws_elasticache_replication_group_test.go

@@ -618,7 +618,7 @@ func TestAccAWSElasticacheReplicationGroup_ClusterMode_UpdateNumNodeGroups_Scale
 	})
 }
 
-func TestAccAWSElasticacheReplicationGroup_ClusterMode_ReplicasPerNodeGroup(t *testing.T) {
+func TestAccAWSElasticacheReplicationGroup_ClusterMode_UpdateReplicasPerNodeGroup(t *testing.T) {
 	var rg elasticache.ReplicationGroup
 	rName := acctest.RandomWithPrefix("tf-acc-test")
 	resourceName := "aws_elasticache_replication_group.test"
@@ -628,6 +628,32 @@ func TestAccAWSElasticacheReplicationGroup_ClusterMode_ReplicasPerNodeGroup(t *t
 		Providers:    testAccProviders,
 		CheckDestroy: testAccCheckAWSElasticacheReplicationDestroy,
 		Steps: []resource.TestStep{
+			{
+				Config: testAccAWSElasticacheReplicationGroupNativeRedisClusterConfig(rName, 2, 1),
+				Check: resource.ComposeTestCheckFunc(
+					testAccCheckAWSElasticacheReplicationGroupExists(resourceName, &rg),
+					resource.TestCheckResourceAttr(resourceName, "cluster_enabled", "true"),
+					resource.TestCheckResourceAttr(resourceName, "parameter_group_name", "default.redis6.x.cluster.on"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.#", "1"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.num_node_groups", "2"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.replicas_per_node_group", "1"),
+					resource.TestCheckResourceAttr(resourceName, "number_cache_clusters", "4"),
+					resource.TestCheckResourceAttr(resourceName, "member_clusters.#", "4"),
+				),
+			},
+			{
+				Config: testAccAWSElasticacheReplicationGroupNativeRedisClusterConfig(rName, 2, 3),
+				Check: resource.ComposeTestCheckFunc(
+					testAccCheckAWSElasticacheReplicationGroupExists(resourceName, &rg),
+					resource.TestCheckResourceAttr(resourceName, "cluster_enabled", "true"),
+					resource.TestCheckResourceAttr(resourceName, "parameter_group_name", "default.redis6.x.cluster.on"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.#", "1"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.num_node_groups", "2"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.replicas_per_node_group", "3"),
+					resource.TestCheckResourceAttr(resourceName, "number_cache_clusters", "8"),
+					resource.TestCheckResourceAttr(resourceName, "member_clusters.#", "8"),
+				),
+			},
 			{
 				Config: testAccAWSElasticacheReplicationGroupNativeRedisClusterConfig(rName, 2, 2),
 				Check: resource.ComposeTestCheckFunc(
@@ -641,11 +667,72 @@ func TestAccAWSElasticacheReplicationGroup_ClusterMode_ReplicasPerNodeGroup(t *t
 					resource.TestCheckResourceAttr(resourceName, "member_clusters.#", "6"),
 				),
 			},
+		},
+	})
+}
+
+func TestAccAWSElasticacheReplicationGroup_ClusterMode_UpdateNumNodeGroupsAndReplicasPerNodeGroup_ScaleUp(t *testing.T) {
+	var rg elasticache.ReplicationGroup
+	rName := acctest.RandomWithPrefix("tf-acc-test")
+	resourceName := "aws_elasticache_replication_group.test"
+
+	resource.ParallelTest(t, resource.TestCase{
+		PreCheck:     func() { testAccPreCheck(t) },
+		Providers:    testAccProviders,
+		CheckDestroy: testAccCheckAWSElasticacheReplicationDestroy,
+		Steps: []resource.TestStep{
 			{
-				ResourceName:            resourceName,
-				ImportState:             true,
-				ImportStateVerify:       true,
-				ImportStateVerifyIgnore: []string{"apply_immediately"},
+				Config: testAccAWSElasticacheReplicationGroupNativeRedisClusterConfig(rName, 2, 1),
+				Check: resource.ComposeTestCheckFunc(
+					testAccCheckAWSElasticacheReplicationGroupExists(resourceName, &rg),
+					resource.TestCheckResourceAttr(resourceName, "cluster_enabled", "true"),
+					resource.TestCheckResourceAttr(resourceName, "parameter_group_name", "default.redis6.x.cluster.on"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.#", "1"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.num_node_groups", "2"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.replicas_per_node_group", "1"),
+					resource.TestCheckResourceAttr(resourceName, "number_cache_clusters", "4"),
+					resource.TestCheckResourceAttr(resourceName, "member_clusters.#", "4"),
+				),
+			},
+			{
+				Config: testAccAWSElasticacheReplicationGroupNativeRedisClusterConfig(rName, 3, 2),
+				Check: resource.ComposeTestCheckFunc(
+					testAccCheckAWSElasticacheReplicationGroupExists(resourceName, &rg),
+					resource.TestCheckResourceAttr(resourceName, "cluster_enabled", "true"),
+					resource.TestCheckResourceAttr(resourceName, "parameter_group_name", "default.redis6.x.cluster.on"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.#", "1"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.num_node_groups", "3"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.replicas_per_node_group", "2"),
+					resource.TestCheckResourceAttr(resourceName, "number_cache_clusters", "9"),
+					resource.TestCheckResourceAttr(resourceName, "member_clusters.#", "9"),
+				),
+			},
+		},
+	})
+}
+
+func TestAccAWSElasticacheReplicationGroup_ClusterMode_UpdateNumNodeGroupsAndReplicasPerNodeGroup_ScaleDown(t *testing.T) {
+	var rg elasticache.ReplicationGroup
+	rName := acctest.RandomWithPrefix("tf-acc-test")
+	resourceName := "aws_elasticache_replication_group.test"
+
+	resource.ParallelTest(t, resource.TestCase{
+		PreCheck:     func() { testAccPreCheck(t) },
+		Providers:    testAccProviders,
+		CheckDestroy: testAccCheckAWSElasticacheReplicationDestroy,
+		Steps: []resource.TestStep{
+			{
+				Config: testAccAWSElasticacheReplicationGroupNativeRedisClusterConfig(rName, 3, 2),
+				Check: resource.ComposeTestCheckFunc(
+					testAccCheckAWSElasticacheReplicationGroupExists(resourceName, &rg),
+					resource.TestCheckResourceAttr(resourceName, "cluster_enabled", "true"),
+					resource.TestCheckResourceAttr(resourceName, "parameter_group_name", "default.redis6.x.cluster.on"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.#", "1"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.num_node_groups", "3"),
+					resource.TestCheckResourceAttr(resourceName, "cluster_mode.0.replicas_per_node_group", "2"),
+					resource.TestCheckResourceAttr(resourceName, "number_cache_clusters", "9"),
+					resource.TestCheckResourceAttr(resourceName, "member_clusters.#", "9"),
+				),
 			},
 			{
 				Config: testAccAWSElasticacheReplicationGroupNativeRedisClusterConfig(rName, 2, 1),

website/docs/r/elasticache_replication_group.html.markdown

@@ -149,7 +149,7 @@ Please note that setting a `snapshot_retention_limit` is not supported on cache.
 
 Cluster Mode (`cluster_mode`) supports the following:
 
-* `replicas_per_node_group` - (Required) Specify the number of replica nodes in each node group. Valid values are 0 to 5. Changing this number will force a new resource.
+* `replicas_per_node_group` - (Required) Specify the number of replica nodes in each node group. Valid values are 0 to 5. Changing this number will trigger an online resizing operation before other settings modifications.
 * `num_node_groups` - (Required) Specify the number of node groups (shards) for this Redis replication group. Changing this number will trigger an online resizing operation before other settings modifications.
 
 ## Attributes Reference