We have configured 1 master and 3 read-only replica databasse in GCP CloudSQL (postgre) for our DHIS2 implementation. We have configure the read replicas in dhis.conf in all the application servers.
However, based on our monitoring, most of the read queries are still directed toward the master database. This created high CPU usage only on our master, and almost no activities in the replicas, thus negating our objective to scale out the database.
Is there any other settings need to be configured in order to use the read replicas more effectively?