I have posted about this before in another thread; unfortunately, I didn’t didn’t get any responses at the time. I’m hoping this forum thread might catch the eye of a Home Assistant “VM expert” or a HA developer that might be able to help effectively troubleshoot this issue after looking at all the clues below:
My Home Assistant VM is intermittently losing TCP connectivity to anywhere outside the VM.
For example, when I try to add the Glances Integration, it can’t even find my Glances server until I try several times. Even after the Glances Integration successfully connects and adds all it’s entities… all the entities will spontaneously drop their connection to the server after a little while …then mysteriously reconnect again. I don’t see any errors/warnings in the Supervisor log. The only way I can get reliable TCP connections to the Glances server is when manually creating REST sensors pointing to Glances API endpoints.. This consistently works always.
So, what’s the difference between the above two methods of connecting to the Glances server? Why do REST sensors (that I manually create) connect reliably …while at the same time, the native Glances Integration’s entities can’t?
Anyway, same thing happens with the MQTT broker addon. it will constantly reconnect to my MQTT clients.
VirtualBox is using Bridged mode; connecting to the HOST’s physical Ethernet adapter (Intel MT 1000 network adapter) with “promiscuous all”. I don’t have any other network adapters and IPV6 is disabled. Pretty simple setup. I don’t have ANY antivirus software (or anything related) running on my HOST (just vanilla Windows 10 with Windows Defender disabled).
It is important to note that I don’t have ANY TCP connectivity issues whatsoever when connecting to my Home Assistant web interface from anywhere. I also don’t have any issues connecting to my Home Assistant’s Samba share or via SSH from anywhere. I also don’t have any issues with manually created REST sensors pointing API endpoints or when pointing to any other web endpoints (such as a cameras, etc) when defined in configuration.yaml.
I would really appreciate any help that might resolve this issue.