Awareness configuration

Elastic Stack Elasticsearch
1 
ElasticSearch Version: 7.17.15
Kibana Version: 7.17.15

Hi, I'm trying to configure the awareness to try to reduce the AWS data transfer cost, but that isn't working. I configured the node.att.zone in each data node using as value the availability zone where the node is located, and in each master node I configured the cluster.routing.allocation.awareness.attributes as zone. The issue is when I increase or decrease the number of replica of the index, in the most case the ELK use an instance in other availability zone. I tried use
cluster.routing.allocation.awareness.force.zone.values, but didn't work too. If someone here can help me, I appreciate.

Data node awareness configuration in elasticsearch.yaml
cluster.routing.allocation.awareness.force.zone.values: cluster_aws_zones
node.attr.zone: date_node_availability_zone

Master node awareness configuration in elasticsearch.yaml
cluster.routing.allocation.awareness.attributes: zone
cluster.routing.allocation.awareness.force.zone.values: cluster_aws_zones
2

Shard allocation awareness is used to ensure that shards are spread out across the zones. This helps the cluster handle the failure of a complete availability zone as shards are spread out and still available.

If you are looking to locate all primaries and replicas within a single availability zone based on an attribute you should instead look at shard allocation filtering. Note that this will leave your cluster susceptible to data loss if a complete availability zone was to fail or become available.