def extract_resource_usage_plot_data(args):
free_json_path = f"{args.prefix}/free_rusage.json"
if not os.path.exists(free_json_path):
free_json_path += ".xz"
if not os.path.exists(free_json_path):
logging.warning(
f"Unable to find memory resource usage data at {free_json_path}.")
return
shadow_json_path = f"{args.prefix}/shadow_rusage.json"
if not os.path.exists(shadow_json_path):
shadow_json_path += ".xz"
if not os.path.exists(shadow_json_path):
logging.warning(
f"Unable to find memory resource usage data at {shadow_json_path}."
)
return
free_data = load_json_data(free_json_path)
shadow_data = load_json_data(shadow_json_path)
__extract_resource_usage(args, free_data, shadow_data)
def __load_user_data(args):
# geographical user info taken from stage command output, i.e., tor metrics
user_data = load_json_data(args.user_info_path)
country_codes = sorted(user_data.keys())
country_probs = [float(user_data[code]) for code in country_codes]
return country_codes, country_probs
def get_relays(args):
data = load_json_data(args.relay_info_path)
relays = data['relays']
stats = data['network_stats']
# sample relays: take all relays that appeared in the input data, and select
# a number that follows the median number of relays that are seen in a consensus.
# this gives us the relays that would represent a full 100% Tor network
sampled_relays, sampled_weights = __sample_relays(relays,
stats['med_count_total'])
# log some info
n_relays = len(sampled_relays['all'])
total_capacity = sum([
relay['bandwidth_capacity']
for relay in sampled_relays['all'].values()
])
gbit = total_capacity * 8.0 / 1000.0 / 1000.0 / 1000.0
logging.info(
"A full Tor network has {} relays with total capacity of {} Gbit/s".
format(n_relays, gbit))
# compute ratios of nodes for each position
pos_ratios = {
'g': len(sampled_relays['g']) / len(sampled_relays['all']),
'e': len(sampled_relays['e']) / len(sampled_relays['all']),
'ge': len(sampled_relays['ge']) / len(sampled_relays['all']),
'm': len(sampled_relays['m']) / len(sampled_relays['all']),
}
# Now that we have a "full" Tor network, scale it down to the requested scale.
n_relays_scaled = round(n_relays * args.network_scale)
chosen_relays, divergence = __choose_relays(n_relays_scaled,
sampled_relays,
sampled_weights, pos_ratios)
relay_count = len(chosen_relays['g']) + len(chosen_relays['e']) + len(
chosen_relays['ge']) + len(chosen_relays['m'])
logging.info("Chose {} of {} relays using scale factor {}".format(
relay_count, n_relays, args.network_scale))
# name the chosen relays
relay_ctr = 1
for pos in ['ge', 'e', 'g', 'm']:
suffix = 'guard' if pos == 'g' else 'exit' if pos == 'e' else 'exitguard' if pos == 'ge' else 'middle'
# use reverse to sort each class from fastest to slowest when assigning the id counter
for (fp, relay) in sorted(chosen_relays[pos].items(),
key=lambda kv: kv[1]['weight'],
reverse=True):
relay['nickname'] = "relay{}{}".format(relay_ctr, suffix)
relay_ctr += 1
return chosen_relays, relay_count
def __load_torperf_datasets(torperf_argset):
torperf_dbs = []
if torperf_argset != None:
for torperf_args in torperf_argset:
torperf_db = {
'dataset': load_json_data(torperf_args[0]) if torperf_args[0] != None else None,
'label': torperf_args[1] if torperf_args[1] != None else "Public Tor",
'color': torperf_args[2],
}
torperf_dbs.append(torperf_db)
return torperf_dbs
def __load_tornet_datasets(args, filename):
tornet_dbs = []
print(args.labels)
label_cycle = cycle(args.labels) if args.labels != None else None
color_cycle = cycle(args.colors) if args.colors != None else None
if args.tornet_collection_path != None:
for collection_dir in args.tornet_collection_path:
tornet_db = {
'dataset': [load_json_data(p) for p in find_matching_files_in_dir(collection_dir, filename)],
'label': next(label_cycle) if label_cycle != None else os.path.basename(collection_dir),
'color': next(color_cycle) if color_cycle != None else None,
}
tornet_dbs.append(tornet_db)
return tornet_dbs
def extract_oniontrace_plot_data(args):
json_path = f"{args.prefix}/oniontrace.analysis.json"
if not os.path.exists(json_path):
json_path += ".xz"
if not os.path.exists(json_path):
logging.warning(
f"Unable to find oniontrace analysis data at {json_path}.")
return
data = load_json_data(json_path)
# parse performance stats only after the network has reached steady state
startts, stopts = args.converge_time, -1 if args.run_time < 0 else args.converge_time + args.run_time
__extract_circuit_build_times(args, data, startts, stopts)
__extract_relay_tput(args, data, startts, stopts)
def extract_oniontrace_plot_data(args):
json_path = f"{args.prefix}/oniontrace.analysis.json"
if not os.path.exists(json_path):
json_path += ".xz"
if not os.path.exists(json_path):
logging.warning(
f"Unable to find oniontrace analysis data at {json_path}.")
return
data = load_json_data(json_path)
# skip the first 20 minutes to allow the network to reach steady state
startts, stopts = 1200, -1
__extract_circuit_build_times(args, data, startts, stopts)
__extract_relay_tput(args, data, startts, stopts)
def extract_tgen_plot_data(args):
json_path = f"{args.prefix}/tgen.analysis.json"
if not os.path.exists(json_path):
json_path += ".xz"
if not os.path.exists(json_path):
logging.warning(f"Unable to find tgen analysis data at {json_path}.")
return
data = load_json_data(json_path)
# skip the first 20 minutes to allow the network to reach steady state
startts, stopts = 1200, -1
__extract_round_trip_time(args, data, startts, stopts)
__extract_download_time(args, data, startts, stopts)
__extract_error_rate(args, data, startts, stopts)
__extract_client_goodput(args, data, startts, stopts)
def extract_tgen_plot_data(args):
json_path = f"{args.prefix}/tgen.analysis.json"
if not os.path.exists(json_path):
json_path += ".xz"
if not os.path.exists(json_path):
logging.warning(f"Unable to find tgen analysis data at {json_path}.")
return
data = load_json_data(json_path)
# parse performance stats only after the network has reached steady state
startts, stopts = args.converge_time, -1 if args.run_time < 0 else args.converge_time + args.run_time
__extract_round_trip_time(args, data, startts, stopts)
__extract_download_time(args, data, startts, stopts)
__extract_error_rate(args, data, startts, stopts)
__extract_client_goodput(args, data, startts, stopts)
