chaosbonobo.py

#!/usr/bin/python

import argparse
import random
import subprocess
import sys
import time

PREFIX = '/sbin/ip netns exec'

DISRUPTIONS = [
(
  '100ms delay',
  'tc qdisc add dev eth0 root netem delay 100ms 50ms 10%',
  'tc qdisc del dev eth0 root netem'
 ),
(
  '1000ms +/- 500ms delay',
  'tc qdisc add dev eth0 root netem delay 1000ms 500ms 25%',
  'tc qdisc del dev eth0 root netem'
),
(
  '1% packet loss',
  'tc qdisc add dev eth0 root netem loss 1%',
  'tc qdisc del dev eth0 root netem'
),
(
  '5% packet loss',
  'tc qdisc add dev eth0 root netem loss 5% 25%',
  'tc qdisc del dev eth0 root netem'
),
(
  '25% packet loss',
  'tc qdisc add dev eth0 root netem loss 25% 25%',
  'tc qdisc del dev eth0 root netem'
),
(
  '100% packet loss',
  'tc qdisc add dev eth0 root netem loss 100%',
  'tc qdisc del dev eth0 root netem'
),
(
  'port blocked',
  'iptables -A OUTPUT -p tcp --dport 8080 -j DROP',
  'iptables -D OUTPUT -p tcp --dport 8080 -j DROP'
),
(
  'port rejecting',
  'iptables -A OUTPUT -p tcp --dport 8080 -j REJECT',
  'iptables -D OUTPUT -p tcp --dport 8080 -j REJECT'
),
]

def exec_disruption(namespace, command):
  subprocess.check_call(PREFIX.split() + [namespace] + command.split())

def main(args):
  running = True
  print '['
  sys.stdout.flush()
  while running:
    disruption = random.choice(DISRUPTIONS)
    disrupt = random.random() <= args.probability
    try:
      if disrupt:
        print '{"name":"%s","tid":0,"pid":0,"ts":%d,"ph":"B"},' % \
            (disruption[0], time.time() * 1000)
        sys.stdout.flush()
        exec_disruption(args.namespace, disruption[1])
      time.sleep(args.period)
    except KeyboardInterrupt:
      running = False
    finally:
      if disrupt:
        print '{"name":"%s","tid":0,"pid":0,"ts":%d,"ph":"E"},' % \
            (disruption[0], time.time() * 1000)
        sys.stdout.flush()
        exec_disruption(args.namespace, disruption[2])
  return 0

if __name__ == '__main__':
  parser = argparse.ArgumentParser(
      description='Simulates bad networking conditions')

  parser.add_argument('namespace', type=str,
                      help='The network namespace')
  parser.add_argument('--probability', type=float, default=0.25,
                      help='The probability in [0.0,1.0) of a disruption')
  parser.add_argument('--period', type=int, default=(5*60),
                      help='The length (in seconds) of a period')

  sys.exit(main(parser.parse_args()))