Hi Roger. The situation nusenu requested the check for sounds similar...
https://lists.torproject.org/pipermail/tor-dev/2018-February/012918.html
Are you saying that in that case dannenberg had Exit in its known-flags? I'm getting kinda tempted to simply drop this check since it seems to be for a weird edge case and I'd rather not continue fiddling with it.
On Thu, Feb 15, 2018 at 11:30 PM, Roger Dingledine arma@mit.edu wrote:
On Thu, Feb 15, 2018 at 10:05:57PM +0000, atagar@torproject.org wrote:
NOTICE: longclaw had 0 Running flags in its vote but the consensus had 6063
Damian: here's another case where we'll want to make the logic smarter.
If longclaw isn't saying it's voting on the Running flag (that is, it doesn't say Running in the "known-flags" line), then it's fine for it to be missing the Running flag in all its networkstatus stanzas.
This situation happens soon after a restart, where the dir auth doesn't think it has a good handle on which relays are reachable and which aren't, so it opts to express no opinion for that round.
--Roger