torps / torps Goto Github PK

after processed server-descriptor file , when I simulate network_state file ,I get the error "ZeroDivisionError: integer division or modulo by zero",I don't know the reason

Hi, I am a senior in university, want to use TorPS to simulate tor path selection algorithm.
While there are some questions may be a bit of infant but that is really I want to know about the tools.

1. Before I use the tool, stem is required, then Is Tor(original code) is also required ?
2. My algorithm use C++ language to realize, Do I have to rewrite my algorithm in Python?
3. I have read the paper roughly - "Users Get Routed - Traffic Correlation on Tor by Realistic Adversaries" , Where can I found more information about it ?

I would like to extend my greatest appreciation if you are so kind as to give me your help ! Thank you very much!

Hello !

The bandwidth weights received from the consensuses don't take into account the bandwidth we add to the simulated network when we inject adversarial nodes. I think that this could have bad consequences, leading to false results when too much bandwidth is added.
Regarding section 3.8.3 of Tor spec, the computation of bandwidth weights depends of the network load.

1. It's possible that, depending of the amount of bandwidth added, the case of the network load change. In results, the bwweights need to be computed differently.
2. Even if the added bandwidth does not modify the case of the network load, the bwweights needs to be recomputed to take into account these adversarial relays in the load balancing.

I plan to modify network_modifiers.py to recompute bwweights. What do you think ?

Regards,

Florentin

Hi Aaron !

https://github.com/torps/torps/blob/master/process_consensuses.py#L21

I think a f.close() is missing at the end of the function ? I didn't set an unlimited number of open file descriptors on my OS and got some errors :-)

Hi, I am trying to run the run_simulations_cat.sh by following the instructions on the main page of torps. It writes
CONGFILE=$BASE_DIR/in/congestion.cator.pickle
at Line 22, but I can't find this file anywhere. Could you please provide this file? I really need it very much to run the simulation.
Thank you!

Hi,
I am trying to run the analyze_and_plot.sh by following the instructions on the main page of torps. I first run the "sh analyze_and_plot.sh" commands, the error is "IndexError: list index out of range", last I find there is no "*.out " and ".time" files, but I don't know where are they from , I checked everything is OK and "stem" is installed on my OS X10.11. Would you please help me? I really need to run the simulator and see the outputs. (I didn't modify any files of torps simulator).
Thank you.

Hi,
I am trying to run the simulator by following the instructions on the main page of torps. I first run the "python process..." commands and it successfully created network state files in the "out" folder. Afterwards, I wanted to run the simulator using "python simulator...." script, but I encounter a strange error and it made me crazy. The error is "AttributeError: 'str' object has no attribute 'bandwidth_weights' ", the error originates from line "consensus.bandwidth_weights" in file "pathsim.py". I checked everything is OK and "stem" is installed on my ubuntu 14.04. Would you please help me? I really need to run the simulator and see the outputs. (I didn't modify any files of torps simulator).
Thank you.

I would like to point out that identifiers like "_TOR_CONTAINER_H" and "__VIVALDINODE_H__" do not fit to the expected naming convention of the C++ language standard.
Would you like to adjust your selection for unique names?

Hi Aaron,

This one is a nasty bug. Currently, the function timestamp (https://github.com/torps/torps/blob/master/pathsim.py#L157) is applied on naive time (without timezone). However, one is received from Stem as a naive UTC (we should report that to Stem, it is a wrong behaviour) and the other is a naive local time. I think that your function will return an offset of the correct timestamp, depending on the local time of the TorPS user.

I think you should enforce using UTC datetime object everywhere. I don't know what are the correctness issues on this but your timestamp function seems to be on the critical path (locate the right desc in process_consensuses, used to convert the published date of the consensus, ect).

A fix would be to:

1. import pytz (pip install pytz)
2. set a tzinfo in timestamp() (td = t - datetime.datetime(1970, 1, 1, tzinfo=pytz.utc)
3. convert all received naive datetime from Stem to timezone-aware utc object (or you will get an exception in timestamp()).

I don't have a solution for 3) but I can search and send you a pull request if you think that everything up makes sense.

Best,
Florentin