AWS::ElastiCache::ReplicationGroup - Sharded cluster always requires replacement when increasing number of nodes

[422158]

Name of the resource

Other

Resource name

AWS::ElastiCache::ReplicationGroup

Description

When you try to increase a number of nodes per shard in AWS::ElastiCache::ReplicationGroup, you always need to replace the resource. It does not matter whether I use ReplicasPerNodeGroup parameter or if I use NodeGroupConfiguration.

1. In case I use ReplicasPerNodeGroup, increase attempt always triggers the replacement.

2. In case I use NodeGroupConfiguration, increase attempt will throw error: Node Group with Id 0001 has replica count 3 (provided that the previous number of nodes in shard 0001 was 3)

Other Details

No response

[aaaeeeo]

+1 I can do this without replacement with ElastiCache directly: https://docs.aws.amazon.com/AmazonElastiCache/latest/red-ug/increase-replica-count.html

[yb-yu]

I need this too. I added a new replica to failover on production, it became impossible to update cloudformation stack.

Is there any way to manually drift before aws supports it?

[ANILGIST]

• 1 waiting for this to be implemented. If we do the changes manually, then CFT would go out of sync.

[expe-elenigen]

Would it be possible to get an update on this, I see there was an official response to this issue saying:

[...] That being said the Cloudformation service team is aware of this behavior change for the ReplicasPerNodeGroup parameter via the console and are working towards releasing a fix so that you can update ReplicasPerNodeGroup via Cloudformation without a replacement of the cluster in the same way you are able to do it via the console. [...]

https://repost.aws/questions/QUsBl1YbQFSiue-TaPdyvtiA/change-the-number-of-replicas-on-my-cluster-without-recreate-it-using-cloudformation

[Aleks-vvt]

+1