We are facing a problem on a customer where (apparently) there are networking issues.
Basically, we have a master DB with several hot_standby slaves, some on the same geo than the master, some on different geo. The application we run uses two JDBC connection pools, one read-only to the local DB replication, one write-only to the master DB.
The odd thing on this case is that the replication process is working, the slave is up to date with the master, but the JDBC connection to the master fails during the handshake process, with a fatal (48) error.
Enabling trust connections on the master for the given host, and disabling SSL validation (&sslfactory=org.postgresql.ssl.NonValidatingFactory) we can connect successfully.
We need a way to debug this issue and understand how the replication connection works ok and the JDBC doesn’t. What next steps do you recommend?
Thanks