Re: [tor-talk] Bursts of 'many circuit creation requests'

[Julian Wissmann <juwi@xxxxxxxxx>]

Just out of curiosity: Do you also get
"[notice] cull_wedged_cpuworkers(): Bug: closing wedged cpuworker. Can
somebody find the bug?
[err] cpuworker_main(): Bug: writing response buf failed. Exiting.
[warn] Tried to establish rendezvous on non-OR or non-edge circuit."
after several days of Tor running?

Julian
> ...
>> Did you reboot your dsl router? IIRC, there was a similar problem
>> when I run a Tor relay behind my Fritz!Box (AVM).
> 
> Nope. The DSL router is NetBSD/sparc, and it doesn't NAT; the tor
> relay machine has its own public IP, so many TCP connections don't
> consume ressources on the router. Earlier occurrences of this were
> dealt with at most a restart of the tor process.
> 
> ....
>>> This is 0.2.3.12-alpha running on a pretty lowly AMD Geode 500
>>> MHz. I've taken the tor node offline for a few more hours.
>> 
>> I don't think that the cpu is the limitting factor. My Tor relay
>> is running on such a machine (Alix-Board) too, (bandwidthrate
>> 64000 B) without problems.
> 
> The problem seems to be that circuit setup requires much more CPU 
> than regular connection processing, and there seems to be no way of
> limiting the number of circuit setups I am willing to do (except 
> that the CPU itself won't process more).
> 
> I'm more interested in why I get smothered in circuits all of a
> sudden.
> 
> *
> 
> One thing I notice: When I open a connection to a hidden service,
> the first thing is a frenzy of circuit building; probably to
> find/reach a rendezvous point. Do I just happen to get elected as a
> rendezvous point for a popular hidden server?
> 
> Andreas
> 

_______________________________________________
tor-talk mailing list
tor-talk@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk