[tor-relays] Consensus Weight Dropping/Authority Issues?

[teor]

Hi John,

>> On 2 Feb 2020, at 14:15, John Ricketts <john at quintex.com> wrote:
>> 
>> I have been watching the consensus weight and bandwidth of all of my 50 exit nodes drop consistently over the past few months. I have not made any hardware changes in my data center and actual
>> customers have not complained about any performance issues.
>> 
>> Operating systems and Tor version are up to date. I'm dedicating a significant portion of bandwidth to these nodes - 10gbit/sec.
>> 
>> Am I having issues with the bandwidth authorities?
>> 
>> I'm growing frustrated with my performance to resources ratio, I should be doing far better than this.
>> 
>> Please throw ideas at me - open to any ideas.
> 
> On Feb 1, 2020, at 20:05, "starlight.2018q2 at binnacle.cx" <starlight.2018q2 at binnacle.cx> wrote:
> 
> The rating shift experienced by your relays and many others is likely a consequence of the gradual phase-in of the SBWS scanner implementation in place of TorFlow.  I for one find SWBS unimpressive.

There are known bugs in both bandwidth authority implementations,
sbws and torflow. Here's a list of the critical sbws bugs:
https://trac.torproject.org/projects/tor/query?status=!closed&keywords=~sbws-majority-blocker

We're working on getting some funding to fix these bugs in sbws.

(Torflow is very old and hard to install, most of its dependencies are
unsupported. So we aren't tracking torflow bugs in detail any more.)

I don't think we've added any sbws instances for a few months. And I
haven't seen any specific evidence that sbws (or torflow) is
responsible for these issues.

If you give us a timeframe, we might be able to help more.

It's also possible that there are more exits in the Tor network, or
that the routing between the bandwidth authorities and your network
has become slower.

There are some other common explanations for slow relays, listed on
this wiki page:
https://trac.torproject.org/projects/tor/wiki/doc/MyRelayIsSlow

You might find some of the tips on that page helpful.

Also, if you can find some authorities that are giving your relays
low measurements, we might be able to find out why.

You can see a copy of all the bandwidth authority measurements here:
https://consensus-health.torproject.org/consensus-health.html

T

--
teor
----------------------------------------------------------------------

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: Message signed with OpenPGP
URL: <http://lists.torproject.org/pipermail/tor-relays/attachments/20200203/bd3a616d/attachment-0001.sig>