CLI can't connect to main process on Windows 10 when using a public network

[RedRanger14]

Bug Report

Summary

On Windows 10 (Confirmed not to be an issue on Windows 11) the app is unable to successfully link with Nexus Mods website to download mods when using a public network

Steps to reproduce

1. On a Windows 10 Machine with a public network (Must restart machine for network changes to take affect)
2. Add a game e.g. Stardew Valley
3. On Nexus mods website download any mod using vortex

The mod will not appear in the Nexus Mods app

What is the expected behaviour?

The mod downloads in the app successfully

Other information

nexusmods.app.main.current.log

Log files can be found in these locations:

• Windows: %LOCALAPPDATA%\NexusMods.App\Logs
• Linux: ~/.local/state/NexusMods.App/Logs

Please attach both nexusmods.app.main and nexusmods.app.slim log files with the most recent dates.

[j-m-harris]

nexusmods.app.slim.07.log

nexusmods.app.main.current.log

[halgari]

Tried this out on a VM that's running Windows 10 Pro v19045.3803. And can't replicate the issue so far. From everything I've reasearched we shouldn't have this issue as long as we are using the loopback interface, which we are using (it's hardcoded to IPAddress.Loopback which links to 127.0.0.1) so we're not even using a host name. We're also using a port of 0 which tells the OS to pick whatever port it wants to listen on.

We'll keep trying to replicate the issue.

[j-m-harris]

To recreate, I've found it's vital to boot with the network in public mode. If you have the app working in private and you switch to public then it continues to work.

For the VM, I was using QEMU with an emulated e1000 NIC.

[halgari]

excellent, giving that at try (rebooting in public mode)

[halgari]

Even with a reboot I'm still not encountering this issue and have been able to download mods and install them. Thinking this was an issue with newer builds (all my previous tests were done on the release from 3 weeks ago), I downloaded a build created today, and also couldn't replicate the issue.

So I'm back to trying to figure out what might be different in the two setups. Can you check the firewall settings @j-m-harris and make sure there's no entries for the app, and/or 127.0.0.1? Also I assume this is a stock configurtion of Windows 10, not a LTS or minimized build?

[j-m-harris]

• I do not see any entries (inbound or outbound) in the firewall settings for the app.
• I do not see any entries (inbound or outbound) in the firewall settings for local address 127.0.0.1 specifically- they are all 'any', apart from fe80::/64 for router advertisment (ICMPv6-Out).

The windows install is stock, Win 10 Pro v19045.5487

I note that although @RedRanger14 could recreate it not working for him (mod installation), he hasn't got the same main app error and has not shared the slim log error. It's possible he's recreated a different problem.

[j-m-harris]

Is the networking perhaps a red herring here? If you follow the logs I posted in #2722 (comment) then it appears the CLI invocation did talk to the main process and sent the NXM URL at 00:01:41.741.

The resulting program invocation then fails soon after at 00:01:42.485 with Encountered an unobserved task exception.

Some 30+ seconds later at 01:18:28.172 the CliServer reports the network problem, but the root cause is whatever failed previously in the task right?