Further to my previous posts…
For anyone who stumbles upon this thread… a clever hack was suggested by Slavisa_Milojkovic for how to deal with this if the Operator does not inject the URI: pg-cluster1-primary into pgbouncer.
You can create a coreDNS rewrite rule that forces requests for the IP for cluster1-primary to to cluster1-ha… See here: Custom DNS Entries For Kubernetes .
For more info, see here: Pgbouncer server DNS lookup failed - PostgreSQL - Percona Community Forum
Not a fix, but a good work around.
I have checked here, on the CrunchyData forum and elsewhere on the internet but there is still no real solution or a reason why cluster1-primary does not have an IP address. Perhaps IP6 has something to do with it?
There is another potential work around… the docs in the Operator repo suggest that we can add databases and users configurations to pgbouncer.ini. but unfortunately it does not indicate what the correct syntax for configuration entries in spec.proxy.pgBouncer.config.databases or spec.proxy.pgBouncer.config.users?
See these posts for more info: Pgbouncer not finding cluster1-primary - PostgreSQL - Percona Community Forum