[tor-talk] Craigslist now giving Tor the slows, lol

[Mirimir]

On 06/11/2014 08:38 PM, grarpamp wrote:
> On Thu, Jun 5, 2014 at 11:52 PM, Mirimir <mirimir at riseup.net> wrote:
>> I've looked at four CL sites:
> 
> Here the distribution of every exit with a matching
> policy to one site, with its respective page time and
> 'getinfo ns/id/fp' weight. Exits sorted by page time.
> 
> http://bayimg.com/nAodGAaFL

Cool. What URLs have you been looking at -- root URLs or particular ads,
posts, etc?

> The first 170+ with 0 time tor refused as 'unrecognized
> relays' even though policy permitted them. They might
> have been sleeping.
> 
> The division of half the relays is interesting.
> If there's a better source for the probability each
> exit will be picked by a client I'd like to know?

You're testing all known exits, right? And they're ranked in the plot by
loading time, right? But in that case, I don't understand the zero
weights in the center.

I'm guessing that Tor clients don't pick at least some of the
non-responding ones. I'd also like a cite to the selection algorithm.

Also, how are you measuring load times? Are you getting full pages with
images etc, or just the HTML?

> As is the sizable number of weight dropouts
> around the start of blocking near 120 sec.

I'd appreciate more explanation of this observation.

>> In the course of each test set, I saw connections through the initial
>> exit to 100-130 other CL sites. I also saw numerous connections to other
>> CL sites through many other exits.
>>
>> I don't know how to interpret that. Any ideas?
> 
> CL DNS load balancing moving IP's, your html agent fetching
> embedded urls.

What I'm wondering is whether Craigslist's distributed design and load
balancing methods are slowing down access via Tor, but not intentionally.