python模块介绍- multi-mechanize 性能测试工具

python模块介绍- multi-mechanize 性能测试工具

2013-09-13 磁针石

#承接软件自动化实施与培训等gtalk：ouyangchongwu#gmail.comqq 37391319

#博客:http://blog.csdn.net/oychw

#版权所有，转载刊登请来函联系

# 深圳测试自动化python项目接单群113938272深圳广州软件测试开发 6089740

#深圳湖南人业务户外群 66250781武冈洞口城步新宁乡情群49494279

#参考资料：python手册

Multi-Mechanize 是一个开源的性能和负载测试框架，它并发运行多个 Python 脚本对网站或者服务生成负载（组合事务)。

Multi-Mechanize最常用于web性能和可扩展性（scalability）测试，也适用于任何python可以访问的API。
               测试输出报告保存为HTML或JMeter的兼容的XML。主要特性：

• 支持各种 HTTP methods
• 高级超链接和HTML表单支持
• 支持 SSL
• 自动处理 Cookies
• 可设置HTTP头
• 自动处理重定向
• 支持代理
• 支持 HTTP 认证

安装

使用标准的python安装方式。注意，需要安装matplotlib以支持作图，在centos6下面可以这样安装yum -y install python27-matplotlib。这里都以linux（centos）为例。

 

 

快速入门：

创建项目：# multimech-newproject my_project

执行项目：# multimech-run my_project

 

 user_groups:  2

 threads: 6

 

[================100%==================]  30s/30s  transactions: 119  timers:119  errors: 0

waiting for all requests tofinish...

 

 

analyzing results...

 

transactions: 125

errors: 0

 

test start: 2013-09-13 11:47:47

test finish: 2013-09-13 11:48:16

 

created:./my_project/results/results_2013.09.13_11.47.46/results.html

 

done.

至此，我们已经完成了一个简单的性能测试。

查看性能测试结果：./my_project/results/results_2013.09.13_11.47.46/results.html

Performance Results Report

Summary

transactions: 125
errors: 0
run time: 30 secs
rampup: 0 secs

test start: 2013-09-1311:47:47
test finish: 2013-09-1311:48:16

time-series interval: 10secs


workload configuration:

group name	threads	script name
user_group-1	3	v_user.py
user_group-2	3	v_user.py

All Transactions

Transaction Response Summary (secs)

count	min	avg	80pct	90pct	95pct	max	stdev
125	1.021	1.499	1.796	1.890	1.972	2.002	0.282

Interval Details (secs)

interval	count	rate	min	avg	80pct	90pct	95pct	max	stdev
1	43	4.30	1.021	1.462	1.688	1.841	1.897	2.002	0.263
2	39	3.90	1.026	1.543	1.830	1.934	1.976	1.986	0.290
3	37	3.70	1.025	1.495	1.806	1.964	1.985	1.988	0.295

Graphs

Response Time: 10 sectime-series

 

Response Time: raw data (allpoints)

 

Throughput: 5 sec time-series

 

---

Custom Timer: Example_Timer

Timer Summary (secs)

count	min	avg	80pct	90pct	95pct	max	stdev
119	1.020	1.497	1.794	1.889	1.970	2.000	0.281

Interval Details (secs)

interval	count	rate	min	avg	80pct	90pct	95pct	max	stdev
1	43	4.30	1.020	1.461	1.686	1.839	1.895	2.000	0.262
2	39	3.90	1.025	1.542	1.828	1.932	1.975	1.984	0.289
3	37	3.70	1.024	1.493	1.804	1.962	1.983	1.986	0.295

Graphs

Response Time: 10 sectime-series

 

Response Time: raw data (allpoints)

 

Throughput: 10 sec time-series

 

---

 

 

目录结构：

 

每个测试项目包含以下内容：

        config.cfg的配置文件。用于设定测试选项。
        test_scripts/虚拟用户脚本的目录。在这里添加您的测试脚本。
        results/:结果存储目录。对于每个测试都声称一个时间戳目录，里面包含结果的报告。

multimech-newproject，默认生成一个随机数的脚本。脚本如下：

# cat v_user.py

 

import random

import time

 

 

class Transaction(object):

    def__init__(self):

       pass

 

    defrun(self):

        r= random.uniform(1, 2)

       time.sleep(r)

       self.custom_timers['Example_Timer'] = r

 

 

if __name__ == '__main__':

    trans= Transaction()

   trans.run()

print trans.custom_timers

 

配置参数的含义如下：

• run_time: duration of test (seconds)
• rampup: duration of user rampup (seconds)
• results_ts_interval: time series interval for results analysis (seconds)
• progress_bar: turn on/off console progress bar during test run
• console_logging: turn on/off logging to stdout
• xml_report: turn on/off xml/jtl report
• results_database: database connection string (optional)
• post_run_script: hook to call a script at test completion (optional)

特别提下 rampup表示多长时间内加载完所有用户。详细的配置参见：http://testutils.org/multi-mechanize/configfile.html

脚本书写：

用Python书写，

测试脚本模拟虚拟用户对网站/服务/ API的请求，

脚本定义了用户事务
            更多内容参见脚本手册：http://testutils.org/multi-mechanize/scripts.html#scripts-label

每个脚本必须实现一个Transaction()类。这个类必须实现一个run()方法。基本的测试脚本结构如下：

class Transaction(object):

    def run(self):

        # do something here

        return

 

运行期间，Transaction()实例化一次，run()方法则反复调用：

class Transaction(object):

 

    def __init__(self):

        # do per-user user setup here

        #this gets called once on user creation

        return

 

    def run(self):

        # do user actions here

        # this gets called repeatedly

        return

从结构上看，如果每次run如果需要setup和teardown，时间也会计算在run里面，会显得事务处理的时间更长。这个就需要使用定时器来精确计时。

另外脚本建议先调试之后在运行，因为Multi-Mechanize有可能报错不够精准。可以这样运行：# python v_suds.py。v_suds.py是你实际使用的脚本名。另外suds这个库好像实现时性能一般，并发200时，客户端cpu占用率经常会100%，为此web service如果要上大量用户的话，建议用其他库替代，比如soapPy。

下例使用mechanize进行web测试。

class Transaction(object):

    def __init__(self):

        pass

 

    def run(self):

        br = mechanize.Browser()

        br.set_handle_robots(False)

        resp =br.open('http://192.168.4.13/env.htm')

        assert (resp.code == 200), 'BadResponse: HTTP %s' % resp.codes

        assert ('service name' inresp.get_data())

 

 

下面用httplib库重写脚本，并增加定时器。通过定时器，可以分析各个步骤的耗时。

import httplib

import time

 

class Transaction(object):

    defrun(self):

       conn = httplib.HTTPConnection('192.168.4.13')

       start = time.time()

       conn.request('GET', '/env.htm')

       request_time = time.time()

       resp = conn.getresponse()

       response_time = time.time()

       conn.close()

       transfer_time = time.time()

 

       self.custom_timers['request sent'] = request_time - start

       self.custom_timers['response received'] = response_time - start

       self.custom_timers['content transferred'] = transfer_time - start

 

       assert (resp.status == 200), 'Bad Response: HTTP %s' % resp.status

 

 

if __name__ == '__main__':

    trans= Transaction()

   trans.run()

 

    fortimer in ('request sent', 'response received', 'content transferred'):

       print '%s: %.5f secs' % (timer, trans.custom_timers[timer])

http://testutils.org/multi-mechanize/scripts.html#scripts-label还有更多的实例：

importmechanize

importtime

 

 

classTransaction(object):

 

    def__init__(self):

        pass

 

    defrun(self):

        # create a Browserinstance

        br = mechanize.Browser()

        # don't bother withrobots.txt

        br.set_handle_robots(False)

        # add a customheader so wikipedia allows our requests

        br.addheaders = [('User-agent', 'Mozilla/5.0Compatible')]

 

        # start the timer

        start_timer= time.time()

        # submit therequest

        resp = br.open('http://www.wikipedia.org/')

        resp.read()

        # stop the timer

        latency = time.time() - start_timer

 

        # store the customtimer

        self.custom_timers['Load_Front_Page'] = latency

 

        # verify responsesare valid

        assert (resp.code ==200), 'Bad Response: HTTP%s'% resp.code

        assert ('Wikipedia, thefree encyclopedia'in resp.get_data())

 

        # think-time

        time.sleep(2)

 

        # select first(zero-based) form on page

        br.select_form(nr=0)

        # set form field

        br.form['search'] ='foo'

 

        # start the timer

        start_timer= time.time()

        # submit the form

        resp = br.submit()

        resp.read()

        # stop the timer

        latency = time.time() - start_timer

 

        # store the customtimer

        self.custom_timers['Search'] = latency

 

        # verify responsesare valid

        assert (resp.code ==200), 'Bad Response: HTTP%s'% resp.code

        assert ('foobar'in resp.get_data()), 'Text AssertionFailed'

 

        # think-time

        time.sleep(2)

importurllib2

importtime

 

classTransaction(object):

    defrun(self):

        start_timer= time.time()

        resp = urllib2.urlopen('http://www.example.com/')

        content = resp.read()

        latency = time.time() - start_timer

 

        self.custom_timers['Example_Homepage'] = latency

 

        assert (resp.code ==200), 'Bad Response: HTTP%s'% resp.code

        assert ('Example Web Page'in content), 'Text AssertionFailed'

importurllib2

importtime

 

classTransaction(object):

    def__init__(self):

        self.custom_timers = {}

        withopen('soap.xml') as f:

            self.soap_body = f.read()

 

    defrun(self):

        req = urllib2.Request(url='http://www.foo.com/service', data=self.soap_body)

        req.add_header('Content-Type', 'application/soap+xml')

        req.add_header('SOAPAction', 'http://www.foo.com/action')

 

        start_timer= time.time()

        resp = urllib2.urlopen(req)

        content = resp.read()

        latency = time.time() - start_timer

 

        self.custom_timers['Example_SOAP_Msg'] = latency

 

        assert (resp.code ==200), 'Bad Response: HTTP%s'% resp.code

        assert ('Example SOAPResponse'in content), 'Text AssertionFailed'

importhttplib

importurllib

importtime

 

 

classTransaction(object):

    def__init__(self):

        self.custom_timers = {}

 

    defrun(self):

        post_body=urllib.urlencode({

            'USERNAME': 'corey',

            'PASSWORD': 'secret',})

        headers = {'Content-type': 'application/x-www-form-urlencoded'}

 

        start_timer= time.time()

        conn = httplib.HTTPConnection('www.example.com')

        conn.request('POST', '/login.cgi', post_body, headers)

        resp = conn.getresponse()

        content = resp.read()

        latency = time.time() - start_timer

 

        self.custom_timers['LOGIN'] = latency

        assert (resp.status ==200), 'Bad Response: HTTP%s'% resp.status

        assert ('Example Web Page'in content), 'Text AssertionFailed'

 

参考资料：

http://testutils.org/multi-mechanize/