I’ve stumbled into an issue I’ve been trying to resolve for a good while now and one that I eventually gave up on sometime in the recent past.
This said though, I’m finally making a thread here in hopes someone more knowledgeable than I will know what the issue might be.
Let me first state that I run quite a few network demanding services and a tor relay seems to be the only one that appears to take my connection down. I’ve tried quite a few fixes but the only solution seems to be disabling the network adapter and re-enabling it.
I’ll summarize all the info with bulletins that way all the information is easy to digest.
-Multiple versions (always latest) of tor used to test – This is on windows 10!
-Happens RANDOMLY (minutes/hour/days) and ONLY when running tor relay
-I use a custom adapter for ethernet (2.5G asus model) with latest drivers
-Guard status or regular relay status does not seem to matter
-I get no issues in the logs for my relay
-Adapter does not overheat or anything of the sort
-I’ve tried running multiple speeds/bandwidth limits on the relay
I have also attached a picture of my ipconfig result which shows the differences between when working and not working.
What kind of bandwidth are we talking? My personal suspicion would be driver related, especially on Windows. Either way your issue sounds like a hard problem to figure out. Worst case, I’d recommend to just automate your hotfix. Create a scheduled task that runs a ps1/batch every x interval or run an agent in your language of choice that simply does what you do manually right now based on a trigger (e.g. network unreachable).
Spikes of 10-20mb/s but nothing surpassing ~30ish.
Setting up an automated hotfix is really not something I’m happy with as I have many SSH sessions running and I would like them to stay alive.
I’m hoping someone can point me in the direction of figuring out exactly what is causing the adapter to behave this way and what tor does that is so different than the rest of the traffic.
What is your relay config like? Are you running exit or just middle/entry?
The amount of bandwidth is not that high to suspect anything there afaik. It could be that Windows doesn’t handle the amount of connections well (i.e. OS/driver). I only use relays on unix based systems, and Windows is not a popular OS to run a relay on.
Would it be an option for you to run the relay inside a VM (Docker/ESXi/Hyper-V)?
But I believe Tor just doesn’t shut down when it can’t handle the amount of simultaneous connections, it just drops some and keeps others alive, am I wrong?
You’re right, it’s not Tor software that I think is at fault. Rather Windows OS behaviour / drivers.
It’s a middle/entry and my config is nothing out of the ordinary. (Meaning no special commands beyond what is required/listed in the guide)
A VM is not out the question (I run unix on the side as well on a laptop/pi) but if at all possible I’d rather not do so because some anti-cheats tend to bitch if you have VM’s running alongside the games.
I suppose I just need to play around with a few drivers of different kinds in hopes it will solve this issue.
Does this mean that you are running the relay on your personal computer/LAN? If so, it is not recommended at all for security reasons. Even it being a middle/entry relay, you may face IP blacklisting and won’t be able to access some services. I’d recommend running a bridge or snowflake instead.
About your issue… I’d try running your relay on a VM running a UNIX-like operating system for a couple days you won’t be gaming, just for testing purposes and see if it drops the connection again, if it does, then you’ll know something’s off with your config or hw set up and not windows. You can save your identity key to a secure place and then set up your relay on Ubuntu, FreeBSD, Fedora… whatever you like the most, this is just to not create a new relay with no consensus.
It is in fact planned to be running on my personal computer. I understand the security risks and would like to proceed anyway.
The blacklisting part I am not afraid of as I have a relay running on my secondary linux system and have been running it for a good length of time with no issues thus far.
I’ll definitely give your suggestions a try and I hope it solves my issue.
I appreciate the time all of you are giving towards the solution here.