RDS instance grafana dashboard / Wrong evaluation for memory usage

[pudovd]

Describe the bug

For the RDS Instance dashboard revision 12 (link) in the Memory panel we can see a lower value than it actually is if there are several RDS instances with the same instance type.

Desktop (please complete the following information)

• OS: Linux
• Prometheus RDS exporter's version: 0.10.0
• RDS engine: PostgreSQL

To Reproduce

We need at least two RDS instances in the same region with the same instance type.

Expected behavior

Current expression is:

100
- sum(rds_freeable_memory_bytes{aws_account_id="000000000000",aws_region="us-east-1",dbidentifier="myrds"}) by (dbidentifier)
  * 100
  / sum(rds_instance_memory_bytes{}
  * on (instance_class) group_left(dbidentifier) rds_instance_info{aws_account_id="000000000000",aws_region="us-east-1",dbidentifier="myrds"}
)
by (dbidentifier)

I think the right expression can be based on the expression for alert from this comment:

100
- max by (aws_account_id, aws_region, dbidentifier) (rds_freeable_memory_bytes{})
  * 100
  / on(aws_account_id, aws_region, dbidentifier) (
      max by (instance_class) (rds_instance_memory_bytes{}) 
      * on (instance_class) group_right() 
      max by (aws_account_id, aws_region, dbidentifier, instance_class) (rds_instance_info{aws_account_id="000000000000",aws_region="us-east-1",dbidentifier="myrds"})
)

Additional context

No addition context.