Mitmdump 0.16 stops working and only passes "clientdisconnect" errors after a couple of hundred requests

I am currently running mitmdump via a bash script (rerun.sh) with following parameters:

gtimeout 30s mitmdump -p 8513 -T --host --anticache -s py.py --ignore :443$’ ./rerun.sh

this is quick and (very) dirty solution because mitmdump stops working and only passes “clientdisconnect” errors after a couple of hundred requests.

this is my py.py:

# This function is run on each response from websites
def response(context, flow):
    with decoded(flow.response):
        if "application/javascript" == flow.response.headers.get('Content-Type', None) and ".js" == flow.request.url[-3:]:
            flow.response.content = flow.response.content + open('js.js', 'r').read()
from libmproxy.models import decoded

and I am getting a lot of these as well, that might or might not be related to the above issue, but perhaps might provide further clues as to what is up:

mitmproxy has crashed!
Please lodge a bug report at: https://github.com/mitmproxy/mitmproxy
192.168.2.7:50851: Traceback (most recent call last):
  File "/usr/local/lib/python2.7/site-packages/libmproxy/proxy/server.py", line 120, in handle
    root_layer()
  File "/usr/local/lib/python2.7/site-packages/libmproxy/proxy/modes/transparent_proxy.py", line 21, in __call__
    layer()
  File "/usr/local/lib/python2.7/site-packages/libmproxy/protocol/tls.py", line 272, in __call__
    layer()
  File "/usr/local/lib/python2.7/site-packages/libmproxy/protocol/http.py", line 130, in __call__
    layer()
  File "/usr/local/lib/python2.7/site-packages/libmproxy/protocol/http.py", line 338, in __call__
    self.get_response_from_server(flow)
  File "/usr/local/lib/python2.7/site-packages/libmproxy/protocol/http.py", line 469, in get_response_from_server
    if flow.response.stream:
AttributeError: 'NoneType' object has no attribute 'response'

while the present workaround is somewhat underperforming it is serviceable for now as i have to deployed this Saturday. But there must be less “hacky” way of solving this issue long term? This is my first time submitting an issue on GitHUb so please bear with me if I am not 100% up to protocol.

Mitmproxy Version: mitmproxy-0.16-osx.tar.gz Operating System: OSX

[KristofferOrum]

I enden up using this Quick and Dirty solution, with a self referring bashscript rerun.sh:

gtimeout 30s mitmdump -p 8513 -T --host --anticache -s py.py --ignore :443$ . /rerun.sh

basically restarting it every 30 seconds - Not a perfect solution by any means, but serviceable in a pinch - Hope it helps…

[ShoutMarket]

Maybe you know how to get the clients ip from inside a script while running mitmdump?