New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Socks5 mode every some minutes drops active network connections outside the dedicated port and clears dns cache. #226
Comments
Well, deleting AG VPN fixed the issue. I'm back to stable network including local: for 9 hours not a single connection has been terminated unexpectedly. Not sure why AG VPN v1.2 sabotages network. |
@deNoor, thank you for reply and sorry for the delay with the answer. Could you please update to the 1.2.2 version we released on March 31 and try to repro the issue again? |
v1.2.2 settings: Cloudflare WARP is present and active in dns only mode. I used 1.2.2 for 3 days, but with Adguard.Vpn.Tools.exe renamed to prevent it being called from AG service. (Log says 'completed successfully' + 'cannot execute due to absence of the file' :D). I also tried to call tools.exe uninstall + flushdns on my own (scripted) every minute for some hours and I got no network interruptions. I had some issues with "connection_reset" errors while trying to access some pages in Chrome (api.veryawesomeprivacy.org as described in this issue) only with Adguard DNS. Those issues were fixed by stopping Adguard VPN service, but didn't happen again after PC reboot with VPN service active. So I'm not sure about impact from AG VPN client. I'll try using v1.2.2 in normal way, with Tools.exe available for service to call. |
Also C:\Windows\System32\drivers\adgvpnnetworkwfpdrv.sys is always running which makes |
I got no connection dropped for 15 hours with v1.2.2. Tools.exe was available (not renamed) and adgvpnnetworkwfpdrv.sys was running. But if I start switching VPN servers (locations) on my own or something else (server side disconnect) causes server switch I still get all active connections interrupted. I mean all connections, even those NOT redirected to local socks5 proxy. |
@deNoor, hello!
After the steps described above please grab the logs from
I hope I answered your questions, if not - feel free to bother me. |
Can't reproduce anymore. :) Not sure if I updated from 1.1.1 to 1.2.2 without reboot or some other actions caused guaranteed "connection reset". When the issue was fixed for me by stopping AG VPN client I also deleted it and rebooted to verify. I haven't faced this issue since reinstallation of AG VPN client v1.2.2. It seems we should treat it as non-reproducible for now.
I'm not sure why you even need to actively detect all applications and guess which one was set to communicate to local port and which one ignores it on purpose. A user is responsible to connect (and reconnect) its apps to local port where socks5 proxy is listening. Why AgVPN client in socks5 mode even cares about something outside the dedicated port? I don't think that you need to forcibly flush dns also. :) Or at least please provide an option to disable any DNS intervention from AG VPN client (like it was before v1.2?). If split-tunneling is the only reasonable option, could you implement split-tunneling option to exclude all applications globally then? Because it would be really hard to add every process. :D At the moment only an app picked from the predefined list can be added, so windows Remote Desktop (mstsc.exe) cannot be added. |
Just tested. Unfortunately, an app added to "split-tunneling" is NOT protected from connection interruption on location switch in socks5 mode. (The app used direct connection and was not redirected to local socks5 proxy obviously.) How to test:
Without AgVPN client intervention this stream never ends on stable connection (at least for many hours). Also checked that in v1.2.2 an app added to "split-tunneling" can be redirected to local socks5 proxy without any issues. So if split tunnelling with option "All Apps to bypass vpn" actually will stop dropping active connections in the future releases this can be a working solution for me. |
First of all, thanks for the detailed answer. Yes indeed we should enhance behavior for the socks5 mode (in part of dropping connections) and for flushing DNS cache. This feature will be released further |
thanks, seems to be a bug, we'll check |
Prerequisites
Issue Details
Expected Behavior
VPN is listening for SOCKS5 port without interfering with any other application.
Actual Behavior
After update to 1.2 I have brief disconnects from local network with all active connections dropped. The dropped connections are not redirected to proxy (e.g. Remote Desktop connections to other PCs in local network, internet radio, direct file transfer, voice chat session) and are not expected to experience any impact from Adguard VPN.
Adguard service log contains entries "Start uninstalling driver via 'C:\Program Files (x86)\AdGuardVpn\Adguard.Vpn.Tools.exe' tool" lines at the time of disconnects.
Log entry are repeated often:
23.03.2022 23:18:44
23.03.2022 23:18:55
23.03.2022 23:19:05
23.03.2022 23:44:03
24.03.2022 02:01:56
24.03.2022 02:02:21
24.03.2022 02:27:19
The result says:
"Uninstalling driver has been completed successfully with the result:
[...]
Executing Adguard.Core.Tools with the args /drv_uninstall has been completed
"
But it will repeat the procedure again soon.
More details
I also tried to execute manually
"C:\Program Files (x86)\AdGuardVpn\Adguard.Vpn.Tools.exe" /drv_uninstall waitForExit=True timeout=120000 elevate=True
and got disconnected from one of two active Remote Desktop sessions. I don't think it's a coincidence, but not every execution causes disconnects and not all connections are dropped which makes me not 100% sure and took 4 days to investigate. :)
Before update to 1.2 my RD sessions where reliably active for week+ in wired local network (from restart to restart). I'll continue testing with AG VPN service uninstalled to make sure it causes the issue.
Questions
The text was updated successfully, but these errors were encountered: