On Fri, 12 Jul 2019 02:23:21 +0000 Alec Larsen hello@alec.ninja wrote:
Hello,
I've only recently joined this list, so I apologise in advance if this is not the appropriate place for my question.
For the past month, I have been operating an exit node ( 89094DFA4158C7A1583EC3A332CDCBC74A28CC0E ) from UnitedIX in Chicago, IL, US. The server has a dedicated gigabit port, and I had hoped to be able to relay around 200 TB of traffic per month, but for some reason my advertised bandwidth has been hovering at just 12 MiB/s since the first few days.
The server itself is an older Supermicro PfSense server with a Xeon E3-1270v3 at 3.5ghz and 32gb of RAM. It is currently running the latest stable release of FreeBSD (12.0p7) and Tor (0.4.0.5).
As far as I can tell, the machine is mostly idle:
- The Tor process is the most active, and I've yet to see it go above 5% CPU.
- Memory usage is under 1.5gb.
- Running `speedtest-cli` (even to different servers using the `--server` option) consistently gives 600mbps+ in both directions to most destinations.
- There's nothing interesting in `dmesg` or the debug logs.
Does anyone have suggestions for what the bottleneck here might be? I'm happy to share more details about my configuration if that would be helpful.
Thank you in advance for any help that you are willing to provide! I think Tor provides a lot of value, and I would like to provide as much bandwidth as I can to the network.
Regardless of any further steps you do, just set up a 2nd instance of Tor right now and let it start warming up. The single most effective way to improve bandwidth efficiency is to add instances. The ideal would be running one for each CPU core, but currently the hard limit is 2 instances per IPv4 address. And it is understandable that getting more public IPv4 is often either not an option or not economically feasible.