[tor-dev] maatuska's bwscanner down since 2017-04-14 -> significant drop in relay traffic

[Tom Ritter]

On 20 April 2017 at 10:09, Ian Goldberg <iang at cs.uwaterloo.ca> wrote:
> On Thu, Apr 20, 2017 at 10:54:21AM -0000, relayoperator at openmailboxbeta.com wrote:
>> Hi Tom!
>> since maatuska's bwscanner is down [1] I see a significant drop of traffic on many of my relays, and I believe this is related.
>> Do you have any update to [2] on when maatuska will report bwscan results again?
>> thanks,
>> a concerned relayoperator
>
> I am also seeing a strange sudden drop in usage:
>
> https://atlas.torproject.org/#details/BCEDF6C193AA687AE471B8A22EBF6BC57C2D285E
>
> What's going on?


I can confirm that maatuska's bwauth has been down since last weekend
and that we've been unable to get the box back up. We're seeing if we
can stand up a new one. No ETA. Sorry for the minimal communication on
it, but haven't had a more detailed update.

To determine if the loss of maatuska's bwauth has been the particular
thing affecting a particular relay is labor intensive at this time:
you'd need to see if maatuska had been on the high-side-of-median for
your relay measured value resulting in a lower assignment when it went
away.  You could do this through historical consensus-health documents
(use the arrow on the detailed page next to your relay) or through
vote+consensus comparisons.  Not user friendly though.

We are working on better bwauth analysis tools in #21992, #21993,
#21994, and #21882 - but those are bwauths-as-a-whole, not individual
relay analysis.

-tom