1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
|
import base64
import functools
import numpy
import os
import psutil
import pytest
import random
import threading
import time
from twisted.internet import threads, reactor
# we have to manually setup the events server in order to be able to signal
# events. This is usually done by the enclosing application using soledad
# client (i.e. bitmask client).
from leap.common.events import server
server.ensure_server()
#
# pytest customizations
#
# mark benchmark tests using their group names (thanks ionelmc! :)
def pytest_collection_modifyitems(items):
for item in items:
bench = item.get_marker("benchmark")
if bench and bench.kwargs.get('group'):
group = bench.kwargs['group']
marker = getattr(pytest.mark, 'benchmark_' + group)
item.add_marker(marker)
#
# benchmark fixtures
#
@pytest.fixture()
def payload():
def generate(size):
random.seed(1337) # same seed to avoid different bench results
payload_bytes = bytearray(random.getrandbits(8) for _ in xrange(size))
# encode as base64 to avoid ascii encode/decode errors
return base64.b64encode(payload_bytes)[:size] # remove b64 overhead
return generate
@pytest.fixture()
def txbenchmark(monitored_benchmark):
def blockOnThread(*args, **kwargs):
return threads.deferToThread(
monitored_benchmark, threads.blockingCallFromThread,
reactor, *args, **kwargs)
return blockOnThread
@pytest.fixture()
def txbenchmark_with_setup(monitored_benchmark_with_setup):
def blockOnThreadWithSetup(setup, f):
def blocking_runner(*args, **kwargs):
return threads.blockingCallFromThread(reactor, f, *args, **kwargs)
def blocking_setup():
args = threads.blockingCallFromThread(reactor, setup)
try:
return tuple(arg for arg in args), {}
except TypeError:
return ((args,), {}) if args else None
def bench():
return monitored_benchmark_with_setup(
blocking_runner, setup=blocking_setup,
rounds=4, warmup_rounds=1)
return threads.deferToThread(bench)
return blockOnThreadWithSetup
#
# resource monitoring
#
class ResourceWatcher(threading.Thread):
sampling_interval = 0.1
def __init__(self):
threading.Thread.__init__(self)
self.process = psutil.Process(os.getpid())
self.running = False
# monitored resources
self.cpu_percent = None
self.memory_samples = []
self.memory_percent = None
def run(self):
self.running = True
self.process.cpu_percent()
while self.running:
sample = self.process.memory_percent(memtype='rss')
self.memory_samples.append(sample)
time.sleep(self.sampling_interval)
def stop(self):
self.running = False
self.join()
# save cpu usage info
self.cpu_percent = self.process.cpu_percent()
# save memory usage info
memory_percent = {
'sampling_interval': self.sampling_interval,
'samples': self.memory_samples,
'stats': {},
}
for stat in 'max', 'min', 'mean', 'std':
fun = getattr(numpy, stat)
memory_percent['stats'][stat] = fun(self.memory_samples)
self.memory_percent = memory_percent
def _monitored_benchmark(benchmark_fixture, benchmark_function,
*args, **kwargs):
# setup resource monitoring
watcher = ResourceWatcher()
watcher.start()
# run benchmarking function
benchmark_function(*args, **kwargs)
# store results
watcher.stop()
benchmark_fixture.extra_info.update({
'cpu_percent': watcher.cpu_percent,
'memory_percent': watcher.memory_percent,
})
def _watch_resources(request):
return request.config.getoption('--watch-resources')
@pytest.fixture
def monitored_benchmark(benchmark, request):
if not _watch_resources(request):
return benchmark
return functools.partial(
_monitored_benchmark, benchmark, benchmark)
@pytest.fixture
def monitored_benchmark_with_setup(benchmark, request):
if not _watch_resources(request):
return benchmark.pedantic
return functools.partial(
_monitored_benchmark, benchmark, benchmark.pedantic)
|
