I have set up splitting read and write operations on a database by following this and this links. This became necessary because I discovered that the database instance regularly clocks 70-100% CPU utilization during operations.
I have not added the custom decorator @frappe.read_only to any function in the codebase yet.
I have also created a CPU utilization log that logs each time the rds CPU exceeds 70% capacity.
However, after splitting read and write operations, I still observe that queries, especially from reports are still routed through the master. Summarily the replica database still sits doing nothing.
My question is how do I confirm for sure that read-write spliting has occurred and what impact does it bring, I do not see any impact on database performance.