[tor-bugs] #23543 [Core Tor/Tor]: prop224: Disconnects on long-lasting HS connections (possibly because of mds)

[Tor Bug Tracker & Wiki]

#23543: prop224: Disconnects on long-lasting HS connections (possibly because of
mds)
-----------------------------------------+---------------------------------
 Reporter:  asn                          |          Owner:  (none)
     Type:  defect                       |         Status:  new
 Priority:  Medium                       |      Milestone:  Tor:
                                         |  0.3.2.x-final
Component:  Core Tor/Tor                 |        Version:
 Severity:  Normal                       |     Resolution:
 Keywords:  tor-hs prop224 prop224-bugs  |  Actual Points:
Parent ID:                               |         Points:  1
 Reviewer:                               |        Sponsor:
-----------------------------------------+---------------------------------

Comment (by asn):

 Replying to [comment:3 arma]:
 > Replying to [ticket:23543 asn]:
 > > {{{
 > > ...
 > > Sep 16 01:13:30.000 [info] connection_edge_reached_eof(): conn (fd 18)
 reached eof. Closing.
 > > }}}
 >
 > Tell me more about this eof? It would appear that your irc client
 decided to hang up.
 >

 Here are the logs surrounding the eof. Seems like it's from natural
 cause...

 {{{
 Sep 16 01:12:56.000 [debug] connection_bucket_refill_helper():
 global_read_bucket now 1073741824.
 Sep 16 01:12:56.000 [debug] connection_bucket_refill_helper():
 or_conn->read_bucket now 1073741824.
 Sep 16 01:13:30.000 [debug] conn_read_callback(): socket 18 wants to read.
 Sep 16 01:13:30.000 [debug] read_to_chunk(): Encountered eof on fd 18
 Sep 16 01:13:30.000 [debug] circuit_consider_stop_edge_reading():
 considering layer_hint->package_window 906
 Sep 16 01:13:30.000 [info] connection_edge_reached_eof(): conn (fd 18)
 reached eof. Closing.
 Sep 16 01:13:30.000 [debug] connection_edge_end(): Sending end on conn (fd
 18).
 Sep 16 01:13:30.000 [debug] relay_send_command_from_edge_(): delivering 3
 cell forward.
 Sep 16 01:13:30.000 [debug] circuit_package_relay_cell(): encrypting a
 layer of the relay cell.
 Sep 16 01:13:30.000 [debug] circuit_package_relay_cell(): encrypting a
 layer of the relay cell.
 Sep 16 01:13:30.000 [debug] circuit_package_relay_cell(): encrypting a
 layer of the relay cell.
 Sep 16 01:13:30.000 [debug] circuit_package_relay_cell(): encrypting a
 layer of the relay cell.
 Sep 16 01:13:30.000 [debug] append_cell_to_circuit_queue(): Made a circuit
 active.
 Sep 16 01:13:30.000 [debug] scheduler_channel_has_waiting_cells(): Channel
 6 at 0x7f8c240de460 went from waiting_for_cells to pending
 Sep 16 01:13:30.000 [debug] conn_close_if_marked(): Cleaning up connection
 (fd 18).
 Sep 16 01:13:30.000 [debug] circuit_detach_stream(): Removing stream 51231
 from circ 4062943041
 Sep 16 01:13:30.000 [debug] connection_remove(): removing socket 18 (type
 Socks), n_conns now 4
 Sep 16 01:13:30.000 [debug] connection_free_(): closing fd 18.
 }}}

 I wonder what's going on here. My IRC client was idle at that time and
 shouldn't have disconnected... IRC clients are supposed to stick around
 for weeks, months, years.

 WRT my IRC client deciding to make a new connection shortly after, I guess
 that's probably normal reconnect behavior from irssi when it detects a
 disconnect.

 I attached some debug logs of the surrounding times. Couldn't find
 something incriminating there...

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/23543#comment:4>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online