I have a pglogical logical replication set from on-prem Postgres 12 to AWS RDS Postgres 12 (this replication has been set up long time ago and was working fine).
I have deployed some breaking DDL changes which required pausing the replication (first to publisher and then to subscriber - a method I was previously using multiple times (successfully). After resuming the replication though, its status is down (and no new data generated on publisher since resuming the replication have been replicated).
When investigating this type of issue on "on premise -> on premise" Postgres replication setup, subscriber's postgres log would contain error (e.g. missing column, unique key violation, missing primary key on replicated table) - in the RDS though, there is nothing.
I have modified the following RDS parameters to given values:
- log_error_verbosity: verbose
- log_replication_commands: 1
- log_min_message: info
- log_min_error_statement: info
None of those helped - still only see ""pglogical apply 34026:1810520736" (PID 6110) exited with exit code 1" - type of messages (there are no any errors on publisher - I only see WAL files accumulating, waiting for replication to catch up).
Does anyone know what config change can I do to see details of logical replication error in RDS log?