On Fri, Aug 19, 2022 at 11:07:18PM +0000, gk@torproject.org wrote:

ERROR: Unable to retrieve the consensus from moria1 (http://128.31.0.39:9131/tor/status-vote/current/consensus): Failed to download from http://128.31.0.39:9131/tor/status-vote/current/consensus (URLError): <urlopen error timed out> ERROR: Unable to retrieve the vote from moria1 (http://128.31.0.39:9131/tor/status-vote/current/authority): Failed to download from http://128.31.0.39:9131/tor/status-vote/current/authority (URLError): <urlopen error timed out>

I believe that something on the internet between hetzner (where consensus-health runs) and MIT has a rule to drop all traffic to/from 128.31.0.39: https://gitlab.torproject.org/tpo/network-health/analysis/-/issues/10

My current guess is that it went into place in January during the DirPort dir auth floods -- somebody saw a lot of flows and blocked the destination and never bothered to unblock it.

I've opened a depictor ticket for a workaround so consensus-health can resume working: https://gitlab.torproject.org/tpo/network-health/depictor/-/issues/17

If somebody here enjoys traceroutes and mtr's and wants to walk me through running them on each host, and they think they can use that output to learn where on the internet this block is in place, let me know. :)

--Roger