On 06/11/2014 02:23 PM, grarpamp wrote: > On Wed, Jun 11, 2014 at 2:42 PM, Mirimir <miri...@riseup.net> wrote: >> Some detect that I'm doing console-level snapshot, and throw errors. >> Some of the errors may be CAPTCHAs. Imgur invites me to apply for a job, >> given that I use the console ;) > >> I prefer using random exits with random timing, because that looks more >> like normal user behavior. What I need is a normal-seeming snapshot >> method, and a way of explicitly flagging CAPTCHA challenges. Once I >> finish working out the kinks, I'll post the scripts somewhere. > > user-agent helps, and you can diff to known good reply, > or allow for % size difference. GUI browsers can often be > driven by console, but their weight will mask fetch timing.
OK, I'll look at that. I'm using Midori, which is a very light WebKit based browser, and it takes PNG snapshots. >> The server that I'm using can handle 10-20 VMs testing in parallel, and >> they can run 24/7 as long as it takes. I could do it faster, but that >> would stand out too much as testing. > > If you're using a top50 list, 1M tor users are likely fetching them > more than you would be. Yes, and I don't want to push it, because that could skew results. I don't use vm for testing, too heavy and extra work. I like VMs for isolating stuff, both tracking and threats. I'm planning to start the test series at some random time (0-2 hr or whatever) after VM boot, and dump archived results somewhere. So I won't need to look at the VM console. Once that's working, I'll just clone the VM 10-20x with random MACs. -- tor-talk mailing list - tor-talk@lists.torproject.org To unsubscribe or change other settings go to https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk
