AWS RDS Aurora Postgres is not hiting cache

[Jose_Miguel_Concha]

We have a chirpstack implementation on an AWS architecture that includes 2 EC2 machines behind a load balancer, using RDS Postgres and Elastic Cache for Redis.

Suddenly the cache usage indicator in the Postgres DB (writer instance) began to decrease, raising CPU usage to almost 100%, increasing the memory used and greatly increasing the number of I/Os billed.

This promotion process stopped when restarting the DB instance and currently remains stabilized, but the cache usage level is still lower than before and the I/O charge is more than 5 times higher.

This occurred without increasing the number of Gateways or connected devices.

I have restarted both instances on RDS and refreshed the EC2 instances with no success. Here are some of the DB monitoring service graphics.

Also here are the DB logs that seems to be related to the issue:

postgresql.log.2024-07-12-1800

Thanks in advance!