[tor-bugs] #16992 [Tor]: "[warn] Failure from drain_fd" in Tor v0.2.7.2-alpha (git-36c0ae6f7834af16) running on Windows XP

[Tor Bug Tracker & Wiki]

#16992: "[warn] Failure from drain_fd" in Tor v0.2.7.2-alpha (git-36c0ae6f7834af16)
running on Windows XP
-------------------------+--------------------------------
     Reporter:  TORques  |      Owner:
         Type:  defect   |     Status:  new
     Priority:  normal   |  Milestone:  Tor: 0.2.7.x-final
    Component:  Tor      |    Version:  Tor: 0.2.7.2-alpha
   Resolution:           |   Keywords:
Actual Points:           |  Parent ID:
       Points:           |
-------------------------+--------------------------------

Comment (by yawning):

 This is rapidly getting to be tech support-ish, but I'll write one more
 reply.

 Replying to [comment:8 TORques]:
 > Thanks for your replay. My router/relay can handle a large number of
 simultaneous connections (tcpsys changed to 256 connections). I just built
 the latest tor from your github and I see the modified message related to
 "Failure from drain_fd"

 I hope the 256 isn't the upper limit.

 Speaking purely in terms of theory, relays that can't handle having 1
 connection open to every single other relay in the Tor network
 simultaneously are bad for the network, since at some point a client will
 request a circuit that spans that pair of relays. If this connection fails
 due to resource limitations (router/os/whatever), that has a impact on the
 client's path selection and hurts anonymity.

 In practice, the actual number is somewhat lower, but I do not have a good
 estimate for it, beyond "I seem to recall hearing a number that was much
 higher than 256 for a busy relay" (See #16997 for "we should collect stats
 on this number").

 > {{{
 > Sep 07 12:34:25.000 [notice] Tor 0.2.7.2-alpha-dev opening new log file.
 > Sep 07 12:34:25.781 [notice] Tor v0.2.7.2-alpha-dev running on Windows
 XP with Libevent 2.0.22-stable, OpenSSL 1.0.2d and Zlib 1.2.8.
 > Sep 07 12:34:27.000 [notice] This version of Tor (0.2.7.2-alpha-dev) is
 newer than any recommended version, according to the directory
 authorities. Recommended versions are:
 0.2.4.23,0.2.4.24,0.2.4.25,0.2.4.26,0.2.4.27,0.2.5.8-rc,0.2.5.9-rc,0.2.5.10,0.2.5.11,0.2.5.12,0.2.6.5-rc,0.2.6.6,0.2.6.7,0.2.6.8,0.2.6.9,0.2.6.10,0.2.7.1-alpha,0.2.7.2-alpha
 > Sep 07 12:34:42.000 [notice] Now checking whether ORPort
 109.103.57.102:9001 and DirPort 109.103.57.102:9030 are reachable... (this
 may take up to 20 minutes -- look for log messages indicating success)
 > Sep 07 12:34:43.000 [notice] Self-testing indicates your DirPort is
 reachable from the outside. Excellent.
 >
 > Sep 07 12:34:43.000 [warn] Failure from drain_fd: No error
 >
 > Sep 07 12:34:43.000 [notice] Self-testing indicates your ORPort is
 reachable from the outside. Excellent. Publishing server descriptor.
 > Sep 07 12:34:47.000 [notice] Performing bandwidth self-test...done.
 > }}}

 So the actual bug is still present? nickm? Is this still something that
 should just be dup-ed?

 > Also, I used bench.exe to see the tor performances on my PC but I have
 no idea if those values are good or bad for a relay. Can you look into and
 write your opinion about?

 *shrugs* Doesn't look like you have AES-NI, and the public key crypto
 operations are an order of magnitude slower than a mobile Haswell i5.
 People do run relays on slower things, so there's nothing wrong with the
 fact that all the crypto is slow, but I wouldn't expect amazing
 performance.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/16992#comment:9>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online