While abusing Tor, I see a fair number of EG fails. I looked briefly at src and man and it seems to hint at CircuitBuildTimeout, which I then set to 90.
But I don't see that initial bump reflected in the below cutoff field. Though I do think I see fewer entries now than in an earlier run. Is CircuitBuildTimeout the right knob? Should I bump NumEntryGuards and auto-distribute load that way? # log Guard EG is failing more circuits than usual. Success 103/151. Use 60/60. 103 completed, 0 unusable, 0 collapsed, and 32 timed out. Timeout cutoff is 60 sec. Guard EG is failing more circuits than usual. Success 123/176. Use 65/65. 123 completed, 0 unusable, 0 collapsed, and 36 timed out. Timeout cutoff is 60 sec. Guard EG is failing more circuits than usual. Success 79/151. Use 71/71. 79 completed, 0 unusable, 0 collapsed, and 48 timed out. Timeout cutoff is 60 sec. Guard EG is failing a very large amount of circuits. Success 81/163. Use 73/73. 81 completed, 0 unusable, 0 collapsed, and 58 timed out. Timeout cutoff is 60 sec. connection_edge_process_relay_cell (at origin) failed. connection_edge_process_relay_cell (at origin) failed. Guard EG is failing a very large amount of circuits. Success 58/151. Use 58/58. 58 completed, 0 unusable, 0 collapsed, and 34 timed out. Timeout cutoff is 60 sec. Guard EG is failing more circuits than usual. Success 122/212. Use 117/117. 122 completed, 0 unusable, 0 collapsed, and 44 timed out. Timeout cutoff is 60 sec. Your network connection speed appears to have changed. Resetting timeout to 90s after 18 timeouts and 1000 buildtimes. No circuits are opened. Relaxed timeout for circuit 119644 (a General-purpose client 4-hop circuit in state doing handshakes with channel state open) to 90000ms. However, it appears the circuit has timed out anyway. 4 guards are live. Short path bias probe response length field (1). -- tor-talk mailing list - [email protected] To unsusbscribe or change other settings go to https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk
