[Tor-BSD] OpenBSD pf rules...

[George Rosamond]

Pete Wright:
> On 12/12/14 11:17 AM, George Rosamond wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA512
>>
>> Libertas:
>>> On 12/12/2014 09:18 AM, George Rosamond wrote:
>>>> I also think it might be worth pinging those people running
>>>> high-bandwidth Tor relays on BSDs, and see what they did.
>>>
>>> That's the thing - it seems that there isn't a single
>>> high-bandwidth OpenBSD Tor relay:
>>>
>>> http://torstatus.blutmagie.de/index.php?Platform=OpenBSD
>>>
>>> It's possible that there's one that hibernates often and I've just
>>> never encountered it in the consensus.
>>
>> Yup.
>>
>> So we have one OpenBSD relay "nycbug1" but it isn't exactly
>> 'high-bandwidth'.
>>
>> Someone with enough OpenBSD familiarity and adequate bandwidth needs
>> to start working on that,  and then document and circulate.
>>
>> On a related note, the high-bandwidth FreeBSD relay operators need to
>> get their tweaks out there also.  It would be very useful to compare
>> notes.
>>
> 
> As you mentioned George it's really hard to do comparisons between
> systems.  I suspect what we are seeing here is that there is a pretty
> small sample rate of *BSD systems running tor as a realy/bridge when
> compared to systems running linux.
> 
> for example - i have two relays, both freebsd.  one is hosted at ARP
> Networks in LA and peers with pretty much everyone at 1 Wilshire.  It
> gets pretty good throughput.  Another system is hosted at my house using
> a 300mbit TimeWarner cable internet bundle.  that one does not get very
> good throughput.  this is despite both systems being identical as they
> are VM's running on mostly identical hardware hypervisors.
> 
> so i suspect one large chunk of the puzzle is how is your network
> carrier viewed on the tor network.  i'm pretty certain TWC is fucking
> with my tor traffic, or giving it super low priority via QoS knobs.  I
> suspect there are many Linux systems running into the same issues, but
> due to the sheer numbers of them there is a higher ratio of high
> throughput nodes.

Great point Pete.

I have a smaller scale example:

I have a number of embedded boxes on residential cable networks.  Lots
of wasted bandwidth in those places, needless to say.

They all received regular traffic as published bridges.  I've noticed a
significant drop in usage a few months ago, to the point of only seeing
occasional usage.

But then again that's more of a general tor-talk topic than Tor-BSD ;)

g