(untitled)

Okay, maybe this only news to me, but still, sharing my experience. So I've been renting a VPS for quite some time. There wasn't much going on there, so I decided why not run a Tor relay and use some of those free resources. Ofc I decided against making it an Exit node due to legal concerns, so it's just a basic relay. It's been running butter smooth for a month or two as a Middle, and at some point network decided that my relay is very good and stable and marked it as a Guard. And ever since then something weird been happening with my VPS: DNS fuarrrkups, network issues, and finally it became completely unresponsive and I had to ask tech support to fix it for me. I didn't realize it was a Tor issue at first, so I restarted everything and because my VPS was completely dead for a day or two, Tor network penalized it and it was downgraded from Guard to a regular Middle relay. Only few days later, when it again became a Guard, I realized the problem was Tor - weird issues came back: networking problems, DNS timeouts, 100% CPU usage, etc. Could it be that glowies try to own as much guards as possible and attack all independent guards? Anyway, I've been running a slightly obsolete version of tor, without hardening, without sandbox mode, I'll try again and let you know how it's going in a month or two.

Try running i2p. Honestly, it's a wonder to me Tor is still as popular as it is with how many eyes it has on it.

>>89681 i2p is so god damn slow, any idea on how to speed it up besides customize the bandwith?

>>89682 The only advice I can give you is install i2pd not the regular i2p router bundle. i2pd is much faster and more streamlined.

>>89683 okay, thanks. I installed i2p++

>>89682 >>89683 Official bundle was very problematic when I tried to use it under VPN from a podman container, but it works very well on the clearnet.

>>89680 OP Tor is probably causing a network Collison with your vpn

>>89686 What do you mean? Just in case, I'm not running it under a VPN, but I have few VPNs and proxy servers running on the VPS.

>>89680 Make sure that you are getting security patches and the latest Tor daemon. The one that Ubuntu ships is often outdated. Also, I would consider running Snowflake relay instead of regular relay. >community.torproject.org/relay/setup/snowflake/ >Could it be that glowies try to own as much guards as possible and attack all independent guards? That is possible but I don't know... I have been thinking about similar cases and it lead me to start worry about the glowies potentially trying to backdoor Systemd or Dbus but I don't have any evidence, though. Why Systemd or Dbus? - Because they are almost universally used.

>>89684 i2pd is the slimmest and best imo. i2p++ Is ok. The dev team is really responsive and chill but I just like i2pd better

>>89689 well the biggest problem i have is thats so slow, otherwise it's nice

>>89690 What is slow? i2pd? i2pd is the fastest fork. i2p is just a slow protocol in general.

>>89680 Simple advice, run Tor daemon with resource limits set. By default, any daemon can suck up all your server's resources. This is normal. ulimit

>mfw I finally get the VPS working >spend hours configuring Tor >it works for a day >ISP blocks it >mfw

>>89681 why not https://yggdrasil-network.github.io/ though? since there, get an ipv6 address you can just use.

>>89695 Use a network bridge dumb ass.

>>89698 Yggdrasil is fast, but doesn't really make a any strong privacy guarantees.

>>89694 What are some good values for tor relay? Does this look good? LimitNOFILE=100000 LimitNPROC=4096 LimitCORE=0

>>89711 looks alright. you could limit the cores to n-1 or n-2 so it stands out if all cores were to go active. the point is that it'll *underutilize* your server if the software isn't malfunctioning. and that your logs won't be as swamped if something goes wrong anyway. 4096 is much less than recommended for a Tor relay so that's good. 25 files / proc looks okay. then you can lift some limits once you have it running slow but comfy. slow and steady wins the race.

And cat /proc/<PID>/limits to view the effective / applied limits

I remember an article by `orignal` where he analyses weaknesses of i2p. It turns out it has zero MiTM protection.