[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #1826 [Tor Relay]: Sending packets after RELAY_END
#1826: Sending packets after RELAY_END
-----------------------+----------------------------------------------------
Reporter: mwenge | Owner:
Type: defect | Status: new
Priority: normal | Milestone: Tor: 0.2.2.x-final
Component: Tor Relay | Version: Tor: unspecified
Keywords: | Parent:
-----------------------+----------------------------------------------------
Comment(by arma):
Interesting. My hope is that the exit relay packages and sends 1000 cells,
then it receives the stream end cells, *then* it receives the extra
sendmes you're talking about. If so, it wouldn't send any more cells for
the streams that it had already ended.
The general problem you describe is known to happen when you're on a slow
link, you click on a web page, and then you click stop soon after. You've
already set in motion the receiving of that page, and your Tor will waste
bandwidth receiving the cells and discarding them since they're for an
unrecognized stream.
We could reduce the impact if we reduce the default circuit package
window, which we can control from the consensus: see
circuit_initial_package_window() in circuitlist.c. But doing so may hurt
overall throughput, just like capping tcp's window size will hurt
throughput.
As Nick says, it would be good to confirm that we're not sending any
further stream cells *after* we get the stream end cell.
I'd expect the cells to be long gone (case 4) in general. I guess for a
really slow exit relay (or a really slow middle hop), that might still be
in queues in the exit relay.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/1826#comment:6>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs