New containers kill NPM proxying

Describe the bug
When I install certain new containers NPM proxying stops working. In the browser you get a 504 gateway time out after about 2 minutes.

This has happened when installing some but not all new containers. Some that cause the problem are:

• Uptime Kuma
• jelly-requests

I've tried checking NPM and docker logs but can't figure out the issue. It's like the requests never get to NPM. When this happens I can still get to all proxied services and NPM directly. Stopping the problem container does not resolve the issue; only removing it does.

Nginx Proxy Manager Version
2.12.6

To Reproduce
Steps to reproduce the behavior:

1. docker compose up new container.
2. Try proxying.
3. 504 time out error

Expected behavior

1. docker compose up new container.
2. Try proxying.
3. Get to proxied site.

fallback_access.log
proxy-host-4_access.log
proxy-host-4_error.log
proxy-host-7_access.log
proxy-host-7_error.log
proxy-host-10_access.log

com.docker.backend.exe.log
httpproxy.log
inference.log
monitor.log

Operating System
Windows 11

Start by reproducing the timeout with Uptime Kuma or jelly-requests on Windows 11, then compare the attached Nginx Proxy Manager access and error logs with com.docker.backend.exe.log, httpproxy.log, inference.log, and monitor.log. Trace why requests stop reaching NPM after the container is created and verify that proxying works without removing the container.