#21969: We're missing descriptors for some of our primary entry guards -----------------------------------+------------------------------------ Reporter: asn | Owner: asn Type: defect | Status: assigned Priority: High | Milestone: Tor: 0.3.1.x-final Component: Core Tor/Tor | Version: Severity: Normal | Resolution: Keywords: tor-guard, tor-bridge | Actual Points: Parent ID: | Points: 1.5 Reviewer: | Sponsor: SponsorU -----------------------------------+------------------------------------
Comment (by asn): Replying to [comment:27 s7r]: > OK getting close. > > So I have manually edited my state file and added a guard I control as first primary guard with `confirmed_idx=0`. It was running just fine with no complaints. At some point, I have configured in torrc of the guard '''PublishServerDescriptor 0''' so that it will stop publishing descriptors. The client instance running the debug branch did not complain instantly because it had a descriptor that was still valid, and the first primary guard while not wanting to publish descriptors any longer was still running and accepting connections. > Thanks for the logs, however I'm not sure if `PublishServerDescriptor 0` is the right methodology to reproduce this bug. Doing that will make your relay do the weird thing of not being in the consensus but still accepting connections. I'm not sure how that behavior influences the guard algorithm exactly and it might complicate the analysis here. I think the easiest test thing to do would be to just shut down your guard relay like normal, and check if the guard is still in the primary guard list when it exits the consensus. -- Ticket URL: <https://trac.torproject.org/projects/tor/ticket/21969#comment:28> Tor Bug Tracker & Wiki <https://trac.torproject.org/> The Tor Project: anonymity online _______________________________________________ tor-bugs mailing list tor-bugs@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs