We can estimate and allocate memory and CPU resources for the keeper pod based on PostgreSQL tuning, but I could not find much on tuning proxy and sentinel.
My assumption would be that they do not need much. The sentinel can probably do with one core or less and 1GB or less. Not sure about the proxy, but likely 1 core and up to 1 GB should work as well.
Please let me know if you have any thoughts on resource allocation for the proxy and the sentinel.