[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

[tor-relays] Re: Relay forest18 is still not on the consensus



Thanks — what you’re seeing matches what we observe.

For operators, there isn’t a simple fix. The most reliable approaches all depend on a fast path back to the EU, where, statistically, most Tor relay traffic (guard, middle, and/or exit) still concentrates.

We see this clearly in our deployments:

EU: 2 × 10 G servers, ~60 guard relays each, ~80 threads, ~512 GB RAM, consistently saturate the full 10 G link.

US East (fastest North America path we could get to EU): 2 × 10 G servers, ~100–200 guards each, ~192–320 threads, ~768–1500 GB RAM, typically only reach ~2–4 G.

US West and South: significantly worse.

In the US, we’re paying for 60–80% unused bandwidth purely to help diversify the network. Running relays outside the EU does help long-term, but early operators bear the cost until there’s enough density to rebalance traffic.

One clarification: the bandwidth authorities measure throughput, not latency directly, through their bandwidth scanners. There are six today (2 US, 1 CA, 3 EU), and they feed their measurements into directory authority votes and consensus weight calculations, as documented by the Tor Project:
https://blog.torproject.org/how-bandwidth-scanners-monitor-tor-network/

For reference, the current bandwidth authorities and their locations are listed here:
https://metrics.1aeo.com/misc/authorities.html

For forest18 specifically, you can easily see the latest bandwidth measurements per authority on the authority vote page here:
https://metrics.1aeo.com/relay/6435C5172690160DA25681BEBA1EA5EAC6F2BE70/#authority-votes
The last column (“Cons Wt”) reflects the throughput observed by the authorities; recent values in the ~1.6–7.9 kb/s range are extremely low, which explains why the relay struggles to gain weight.

Also worth noting: despite the challenges you've shared, your relays currently rank #2 on our Frontier Builders leaderboard, which tracks operators pushing capacity into underrepresented regions:
https://metrics.1aeo.com/#frontier_builders
That contribution really does matter!

—
Tor at 1AEO


On Monday, December 29th, 2025 at 12:57 AM, forest-relay-contact--- via tor-relays <tor-relays@xxxxxxxxxxxxxxxxxxxx> wrote:

> 

> 

> Hello.
> 

> Tor at 1AEO wrote:
> 

> > Currently, looks like two main issues and a minor third issue for
> > forest18
> 

> 

> Right now the biggest issue is that, although they unblocked the IPs not
> too long ago, they've been re-blocked and they are now "looking into the
> issue". Hopefully they will unblock them again.
> 

> > 2) Fast flag: The speed is also very slow, shown by very low bandwidth
> > scanner results from the 6 bandwidth directory authorities (Consensus
> > Weight from Authorities, i.e. "Cons Wt" in source below).
> 

> 

> I'm having that problem with many of my relays. The relay's CPU and port
> are absolutely capable of handling a sustained 100 Mbps. From what I
> understand, the problem is that the bandwidth authorities are mostly
> concentrated in the Americas and western Europe, so they mistake high
> latency for poor bandwidth capacity. I don't know how to fix that. I
> assume running multiple relays on the same VPS would help, although that
> seems to me like a crappy hack to patch up a real problem.
> 

> Meanwhile, of the 33 VPSes I have, each of the 3 in the Netherlands that
> I bought on a Black Friday sale simply because they were cheap pass more
> traffic than all my non-US, non-EU servers combined, despite scoring a
> bit low on speedtests. It would be nice if authorities simply increased
> the consensus weight of distant servers proportional to their distance.
> 

> It's a bit discouraging to run a relay in South Africa for months and
> have it pass 0.4 MiB/s when a server in Europe for a fraction of the
> price and a slower network port and CPU passes 12 MiB/s after just a few
> weeks.
> 

> > 3) Uptime: Relay reports and operator directly controls, also shows
> > low, ~2 days.
> 

> 

> I brought all my servers down for a few hours recently. That could be
> the reason for that.

Attachment: image.png
Description: PNG image

Attachment: publickey - tor@1aeo.com - 0x9288289B.asc
Description: application/pgp-keys

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
tor-relays mailing list -- tor-relays@xxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to tor-relays-leave@xxxxxxxxxxxxxxxxxxxx