Database replication stopped on geo testbed

Here is a list of items I would like to complete to help make the replication on the testbed more stable:

• Ensure ssl configuration is chef managed
• Turn off pgbouncer service on geo1 db nodes
• Remove dedicated pgbouncer node for geo2
• Enable FDW - https://gitlab.com/gitlab-com/infrastructure/issues/3389#note_51469322
• Make the secondary part of the repmgr cluster
• Turn off automatic failovers so the Geo secondary doesn't become master