[tor-relays] Rampup speed of Exit relay

[teor]

> On 22 Sep 2016, at 04:40, D. S. Ljungmark <spider at takeit.se> wrote:
> 
> On tor, 2016-09-22 at 12:08 +0200, Aeris wrote:
>>> 
>>> Scaling up on more hardware is always an option, but I really want
>>> to
>>> push the limit of the exit node, as the others won't be exits
>>> (Local
>>> network design, really) , and exit traffic is always more
>>> interesting.
>> 
>> When I say another instance, it’s on the same hardware.
>> Because Tor is not fully multi-thread/multi-core, you have to run
>> another Tor
>> daemon on the same host to use 1 more CPU core and so drain another
>> 150-300Mbps.
>> Currenly, you can start up to 2 Tor daemons per IP, there is a
>> limitation to
>> avoid Sybil attack.
>> 
>> Regards,
> 
> 
> Yes, I'm aware of this, but if I can't get Tor to scale up to even
> 300Mbps on a single instance, adding another instance on the same
> hardware isn't going to magically make it reach saturation. It might
> improve things, seen from the network scaling it from 120Mbps to
> 250Mbps, but it's certainly not going to push it to 700Mbps.

It could - I started the exits radia0 and radia1 about a month ago, at almost the same time, with the same config (different IPs and ports), on the same gigabit link, with plenty of processor. radia0 (350 Mbps) is pushing almost double that of radia1 (200 Mbps):

https://atlas.torproject.org/#search/radia

> First, I want to find _why_ I'm not peaking either CPU usage, or
> bandwidth usage. That the network is a stochastic and semi-random
> limiter, I'm quite aware of, and adding more resources to pool that up
> is doable.

Please see my other email for details of how to measure your capacity locally.

That said, these random factors are genuinely random, and have quite a wide range of results.
We're working on measuring relay capacity better.

> ( Though I can't add more ipv4 addresses right now, ipv6 is plentiful,
> sadly, tor doesn't do very well on ipv6. )

You can run two tor instances per IPv4 address.

Alternately, if your relay fingerprint has had a particularly bad set of random selections, you could try deleting the RSA and Ed25519 identity keys, and starting again. This will reset your reputation on the network.

But that's effectively like starting another Tor instance, which you can already do without destroying your existing one.

Tim

Tim Wilson-Brown (teor)

teor2345 at gmail dot com
PGP C855 6CED 5D90 A0C5 29F6 4D43 450C BA7F 968F 094B
ricochet:ekmygaiu4rzgsk6n
xmpp: teor at torproject dot org






-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.torproject.org/pipermail/tor-relays/attachments/20160922/6a0c28a9/attachment.sig>