Skip to content

Latest commit

 

History

History
85 lines (72 loc) · 3.72 KB

asyncio_readme.md

File metadata and controls

85 lines (72 loc) · 3.72 KB
Raw

#Moving towards scalability...

  1. Have a full backup - database, old files. Make double sure you have it already.
  2. Ensure Python 3.9 venv is present (apt install python3.9-venv python3.9-dev)
  3. python3.9 -m venv ~/venv/
  4. source ~/venv/bin/activate
  5. pip3 install -r requirements.txt --upgrade
  6. If you are using QuestEnhancer you also need async PD - just just download PogoDroid_async.apk from usual place.

###Modes of operation:

  • Single Host (small setups, approx <20)

    Use start.py like you normally would after having installed requirements.txt

  • Single MAD core + MitmMapper in a single process (1 core, small to mediocre setups), multiple MitmReceiver hosts/processes anywhere
    • Use start.py like normally
    • Start additional MitmReceivers using start_mitmreceiver.py
    • Check config parameters as to how to connect additiona MitmReceivers using gRPC to the main MAD instance (namely mappingmanager_ip and mitmmapper_ip need to point to main MAD host)
    • Adjust LoadBalancer behaviour (or Reverse Proxy) accordingly.
      1. Set MitmReceiver's unix sockets per process using mitm_unix_socket with full paths to the socket files
      2. Point ReverseProxy/LoadBalancer to those accordingly Sample NGINX + Supervisord: NGINX config:
      upstream mad_mitm_receiver {
  server unix:/tmp/mad_mitm_receiver_1.sock fail_timeout=0 max_conns=100 max_fails=0;
  server unix:/tmp/mad_mitm_receiver_2.sock fail_timeout=0 max_conns=100 max_fails=0;
}

server {
      listen 80 default_server;
      listen [::]:80 default_server;
      root /var/www/html;

      # Add index.php to the list if you are using PHP
      index index.html index.htm index.nginx-debian.html;

      server_name _;

      location / {
              proxy_set_header Host $http_host;
              proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
              proxy_redirect off;
              proxy_buffering off;
              proxy_pass http://mad_mitm_receiver;
      }

}
      Supervisord config: 
      [program:mad_mitm_receiver]
numprocs = 2
numprocs_start = 1
process_name = mad_mitm_receiver_%(process_num)s
command=/opt/MAD/venv/bin/python3 /opt/MAD/start_mitmreceiver.py -os -vv --mitm_unix_socket=/tmp/mad_mitm_receiver_%(process_num)s.sock
environment=PATH="/opt/MAD/venv/bin:%(ENV_PATH)s"
user = www-data
directory=/opt/MAD
#directory=/opt/map_all/MADDev
startsec=15
startretries=3
autorestart=true
stopsignal=INT
stopwaitsecs=60
stdout_logfile=/var/log/mad_mitm_receiver.log
redirect_stderr = true
  • MAD Core, MitmMapper running on same host (or each on their own) + multiple MitmReceivers as shown above
    • Use start_core.py, start_mitmmapper.py, start_mitmreceiver.py accordingly
    • Adjust mappingmanager_ip and mitmmapper_ip in the configs

Please provide feedback on runtype and performance, especially amount of devices being run in a single MAD core instance :)

Keep in mind that all of MAD is now based on asyncio. Meaning a single .py script being executed runs all code in a single core.

Thus MitmReceiver may show high CPU usage (on a single core). A lot of time is spent waiting for DB processing as well.

Also, the default db_poolsize of 2 may be too little in some situations, do play around with that parameter.

Potential performance improvement:

pip3 install uvloop

TODO:

  • Consider splitting up more components, i.e. MappingManager to be extended to gRPC entirely
  • Split RouteManagers to components
  • Thus connect multiple WebsocketServers to scale by device/worker count