Azure Database for PostgreSQL: Logical Replication Slot Synchronization Issue After HA Deployment

Question

Problem description

I am using Azure Database for PostgreSQL flexible server in the Central US region with PostgreSQL 16.15, configured for zone-redundant high availability. I have set up logical replication for a downstream CDC consumer using the pgoutput plugin. Despite the server reporting as healthy and HA functioning correctly, the synchronization metric for my logical replication slot has remained at 0 for approximately 14 hours. Production traffic and CDC operations are ongoing, but I am concerned about the lack of confirmation that the replication slot has synchronized after failover.

Environment

Azure Database for PostgreSQL flexible server, Central US, PostgreSQL 16.15, SKU: GP_Standard_D8ds_v5, Storage: 1 TiB, zone-redundant HA, primary zone: 3, standby zone: 1, logical replication via pgoutput.

What I've already tried

I reviewed the Azure documentation on logical replication and HA, confirming that PostgreSQL 16 supports slot preservation across failover when configured correctly. I verified that 'hot_standby_feedback' is enabled, checked the slot synchronization metric in Monitoring over the past 24 hours, and reviewed the PostgreSQL logs around the restart window for any replication or slot-sync errors. I also confirmed that the logical replication slot existed before enabling HA and that no slot recreation has been performed, as that is not supported at this time.

Current status

Currently, the HA status remains healthy, and production traffic along with CDC is functioning normally. However, the logical replication slot's sync metric has remained at 0 for about 14 hours. I seek guidance on how to diagnose why the slot is not synchronized and whether there are additional configuration steps or log investigations needed to resolve this issue.