Discovery mechanism for Primary/Replica

[schmitch]

I know that enterprise has a feature like that however only for aurora.

However it would be great if it would support a way to read postgres databases from some kind of store like Kubernetes/Patroni.
Currently we use postgres-operator and while it has two services for primary/replica it does group all the replicas into one Service which is probably not the best way to use it directly with pgdog?

[levkk]

Hi!

Is your pg self-hosted with Patroni/Kube or are you on AWS RDS/GCP SQL/etc.?

We want to add support for RDS Postgres "autodiscovery" by loading the DBs directly from the AWS API. The Aurora one was added first because we could just use aurora_replica_status() directly, which was reasonably quick. We could add support for Patroni too, if that's what you're using as a control plane.

Generally speaking, we are trying to make any convenience, "operating PgDog at scale", features enterprise so people have a reason to pay us 😆. Maybe one such feature by itself may not be worth the procurement effort or the price tag, but hopefully over time, we build enough of them for people to click the Buy button.

it does group all the replicas into one Service which is probably not the best way to use it directly with pgdog?

That's right. You want PgDog to do the load balancing, so direct connection to each replica.