We have an AWS RDS for PostgreSQL on 13.15 and AWS Redshift gets data from the instance using AWS DMS. To upgrade the 13.15 to 16, we are testing with Blue/Green approach. It is necessary to stop the DMS jobs and drop the logical replication slots to carry out a Blue/Green deployment. That part works fine. After switchover, I created a logical replication slot with the same name/type as the one I dropped, on the new RDS instance. But the DMS job fails with errors. In short, the question is - if we stop a DMS job, drop the replication slot, the RDS instance is upgraded, how can we make the DMS job take over from where it left off? The error is
SQL_ERROR SqlState: 08001 NativeError: 101 Message: FATAL: no pg_hba.conf entry for host ""10.*.0.*"", user ""x"", database ""x"", no encryption [1022502] (ar_odbc_conn.c:604)"
But I think that error message is misleading.
It was indeed a connectivity issue. The parameter rds.force_ssl was 0 in RDS 13.15 and the default value in Version 16 is 1. Changing it to 0 resolved the connectivity issue.