Dockerized RabbitMQ based Windows App stops consuming from TCP connection after 20-30 minutes idle

The dockerized application starts running fine, no issues, all messages recieved by all containers.
After a period of idleness my Docker container A stops receiving messages from a TCP connection. Even though there are established connections as seen by the netstat -a in the container A, the published messages from the other Docker container B stop being picked up by the Docker container A.
The connection is initially established from the Docker container A to the Docker host listening on port 5672(RabbitMQ server default port).
That same setup with RabbitMQ service on Docker host is working just fine even after a whole day of idleness, when the app is running on bare metal, it only stops consuming message while running in Docker Desktop.
There is no official RabbitMQ Windows image, so I can’t run it in the same network as the other containers.

I have tried setting the vpnKitMaxPortIdleTime to 0, as mentioned in some web articles, but it didn’t help.

My Host OS is: Windows 10 Professional x64 v1909 build 18363
Docker Desktop is: 4.19.0, Engine: 23.0.5
Base image for my dockerized Windows processes is Microsoft Artifact Registry
The Hyper-V virtual switch is the default Docker Desktop’s NAT switch

This was related to my consumer services connecting and listening to external IP of the host workstation(rabbitmq host). When I switched to using host.docker.internal as the rabbitmq host the timeouts stopped happening.