RDS aurora postgres db.r6i changing network_type from IPV4 to DUAL is not possible with confusing error message

Question

Hello!

We have 2 clusters, running in the same region:
1) Cluster1, aurora postgresql 13.20.3, db.r6i.large
2) Cluster 2, aurora postgresql 13.20.3, db.t3.large

Both clusters are located in same subnets with ipv6 cidrs assigned.

When I try to change `network_type` parameter (terraform) from IPV4 to DUAL on Cluster1, error is occured:
```
api error InvalidParameterCombination: aurora-postgresql 13.20 engine version is not supported for modifying clusters to DUAL network type.
```
Also network_type option is not available via Aws Console for Cluster1.

On the other hand, changing network_type (both terraform & aws console) is available for Cluster2, which is also 13.20.3, but family is db.t3.

Questions:
1) Does it mean that error message corresponds to family (or to family+engine combination) rather than to engine version?
2) If yes, are there any references in documentation about these limitations and workarounds?

re:Post Agent · Answer

The error message you're encountering is indeed confusing as it suggests the engine version is the limitation, but the actual limitation appears to be related to the instance class (db.r6i.large vs db.t3.large) rather than the PostgreSQL version.

Based on your experience with the two clusters running the same Aurora PostgreSQL version (13.20.3) but different instance classes, it appears that the network type modification from IPV4 to DUAL is supported on certain instance families but not others. In this case, the t3 family supports the change while the r6i family does not.

The NetworkType parameter is used to determine whether a DB cluster supports IPv4 protocol only or both IPv4 and IPv6 protocols (DUAL). This network type is determined by the DBSubnetGroup specified for the DB cluster.

Unfortunately, there isn't explicit documentation that lists which instance families support changing to DUAL network type for specific Aurora PostgreSQL versions. The error message is misleading as it suggests the limitation is with the engine version, when your experience demonstrates it's likely related to the instance class.

As a workaround, you might consider:
1. Creating a snapshot of your db.r6i.large cluster and restoring it to a supported instance class
2. Performing a major version upgrade if a newer PostgreSQL version supports DUAL network type on r6i instances
3. Migrating to a different instance class that supports DUAL network type (like your t3 instance)

When making such changes, remember that you may need to reboot the database after the modification completes to apply all parameter changes.  
**Sources**  
[ModifyDBCluster - Amazon Relational Database Service](https://docs.aws.amazon.com/AmazonRDS/latest/APIReference/API_ModifyDBCluster.html)  
[Working with a DB cluster in a VPC - Amazon Aurora](https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/USER_VPC.WorkingWithRDSInstanceinaVPC.html)

Balaji T P · Answer

The network_type parameter in Amazon Aurora is not just about the instance type or engine version — it's determined by the DB subnet group.

If your DB subnet group does not include subnets with IPv6 CIDR blocks, then setting network_type = "DUAL" will fail, even if the instance type and engine version technically support it.
This is exactly what happened in your case with Cluster1.

Create a VPC for dual-stack mode 
==========================

* Creating a VPC with IPv6 support
* Creating subnets with IPv6 CIDR blocks
* Creating a DB subnet group using those subnets
* Launching an Aurora cluster with network_type = "DUAL"

https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/CHAP_Tutorials.CreateVPCDualStack.html

RDS aurora postgres db.r6i changing network_type from IPV4 to DUAL is not possible with confusing error message

Create a VPC for dual-stack mode

Add your answer

Relevant content