Facing issue while migrating azure postgres flexi server to GCP

We are currently migrating data from an Azure Database for PostgreSQL – Flexible Server instance to Google Cloud PostgreSQL using Google’s Database Migration Service (DMS). A HA VPN tunnel has been successfully established between Azure and GCP, with BGP configured to dynamically exchange routes between both networks. The migration initiates without issue, and approximately 90 GB of data is transferred successfully. However, after this point, the DMS job consistently fails due to the BGP session going down. From Azure’s perspective, this indicates that the Azure Virtual Network Gateway is losing connectivity with the BGP peer on the GCP side, resulting in the withdrawal of learned routes and disruption of the VPN tunnel. This could be caused by an issue with BGP keepalive or hold timers, session stability under sustained load, or possible throttling or packet loss at the VPN gateway. We need assistance in identifying whether the Azure VPN Gateway is encountering any errors or resource constraints during the high-volume data transfer, and whether there are logs or diagnostics indicating why the BGP session is being terminated unexpectedly after prolonged traffic flow.