[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-bugs] #20423 [Core Tor/Tor]: Warning notice log: assign_to_cpuworker failed. Ignoring.

#20423: Warning notice log: assign_to_cpuworker failed. Ignoring.
--------------------------------+------------------------------------
 Reporter:  Felixix             |          Owner:
     Type:  defect              |         Status:  new
 Priority:  High                |      Milestone:  Tor: 0.2.9.x-final
Component:  Core Tor/Tor        |        Version:  Tor: 0.2.8.9
 Severity:  Normal              |     Resolution:
 Keywords:  regression freebsd  |  Actual Points:
Parent ID:                      |         Points:
 Reviewer:                      |        Sponsor:
--------------------------------+------------------------------------

Comment (by Felixix):

 Replying to [comment:3 nickm]:
 > I think that the root problem here is whatever causes Tor to stall and
 overload, eating a bunch of CPU.  That would cause all of:
 >     * The "clock jumped X seconds forward" warning
 >     * The remote side to close the circuit and its channel already
 because we took too long...
 >     * ...leading to the duplicate close warning when we try to close it
 for being too old...
 >     * ...and/or leading to assign_to_cpuworker() failing because the
 channel isn't there any more.

 The other way around. Tor is stalling for about X seconds at 5% CPU. And
 58 minutes back at 70-90%. Not the typical overload. I think no circuits
 are handled then.


 > Did you see "Your system clock just jumped X seconds forward" warnings
 with older versions of Tor?  If not, we should try to figure out why those
 started.

 I never had those warnings before (2.6.9 - 2.7.6, see ticket text). Roll-
 back from 2.8.9 (like 2.8.7 before) to 2.7.6 and warnings and X seconds
 are gone.

 > Additionally, we should clean up the warnings for assign_to_cpuworker()
 failing (that shouldn't be a warning), and the duplicate close warning.
 (There should be a check for whether the circuit is marked in the loop at
 the end of onion_pending_add().)

 I am glad I received warnings. Otherwise I might not have noticed it.

 Thanks and cheers

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/20423#comment:4>
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