python下Pool与target方法写在同一个类里要注意的坑-创新互联

在工作中遇到要对开发的接口做压力测试，以前没有做过开清楚什么压测工具好用，正好接口不是什么复杂的接口，curl -X post "接口地址" --data-binary @二进制认证文件 OK！（@表示验证数据是文件类型）

既然这样那我就写个脚本好了,脚本内容如下:

======================================================================
#!/usr/bin/evn python
#_coding:utf8_
from multiprocessing import Pool,Queue
import time,subprocess,os
class YaCe(object):
def init(self,api,binfile,maxpool,qu,maxrequest=100000,status="success"):
self.api = api
self.binfile = binfile
self.status = status
self.maxpool = maxpool
self.maxrequest = maxrequest
self.qu = qu
def prorequest(self):
for i in range(self.maxrequest):
self.qu.put(i)
print(i)
for i in range(int(self.maxpool)):
self.qu.put(None)
print("None")

def conumers(self,i):
    while True:
        data = self.qu.get(True)
        if data == None:
            print("进程%s任务完成..."%i)
            break
        else:
            command = subprocess.getoutput("time curl -X POST --connect-timeout 10 '%s'  --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
            if self.status == "success":
                logfile = os.getcwd()+"/"+"headbeat.log"+"_%s"%date_time
                if "CgoyMDAwMDAwMDAw" in command:
                    print("进程%s__%s..."%(str(i),str(data)))
                    with open(logfile,"a") as f:
                        f.write(command+"\n")
                    f.close()
                else:
                                        print("进程%s__%s..."%(str(i),str(data)))
                    with open(logfile,"a") as f:
                        f.write("Faild\n")
                        f.write(command+"\n")
                    f.close()
            else:
                logfile = os.getcwd()+"/"+"roomlist.log"+"_%s"%date_time
                #print("time curl -X POST '%s'  --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
                command = subprocess.getoutput("time curl -X POST '%s' --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
                if "CAES+" in command:
                    print("进程%s__%s..."%(str(i),str(data)))
                    info = command.split('\n')[-3:]
                    info1 = "\n".join(info)
                    with open(logfile,"a") as f:
                        f.write(info1+"\n")
                    f.close()
                else:
                    print("进程%s__%s..."%(str(i),str(data)))
                    with open(logfile,"a") as f:
                        f.write("Faild\n")
                        f.write(command+"\n")
                    f.close()
def multirun(self):
    ps = int(int(self.maxpool) - 1)
    p = Pool(ps)
    for i in range(self.maxpool):
        print("开启子进程%s"%i)
        p.apply_async(self.conumers,args=(self,i))
    print('等待所有添加的进程运行完毕。。。')
    p.close()
    p.join()
    endtime = time.strftime("%Y%m%d_%X",time.localtime())
    if self.status == "success":
        logfile = os.getcwd()+"/"+"headbeat.log"+"_%s"%date_time
    else:
        logfile = os.getcwd() + "/" + "roomlist.log"+"_%s"%date_time
    with open(logfile,"a") as f:
        f.write("============[%s]============\n"%endtime)
    f.close()
    print('End!!,PID:%s'% os.getpid())

if name == "main":
q = Queue()
Yc = YaCe('压测接口','二进制证认文件',开多少个进程,queue(队列),maxrequest=100（模拟测试多少次访问）,status="faild"（这里因为测试的两个接口，返回不一样用status参数区分测试的接口的返回值处理）)
Yc.prorequest()
print("++++++")
global date_time
datetime = time.strftime("%Y%m%d%X",time.localtime())
Yc.multirun()

====================================================================================
问题
到这里写完了，测试的问题来了，从脚本来看如果运行成功，会有多进程在处理队列的输出，可是结果的输出确是如下：

01
2
3
4
5
6
7
8
9
None
None
++++++
开启子进程0
开启子进程1
等待所有添加的进程运行完毕。。。
End!!,PID:4819

原因
子进程conumers方法完全没有运行，也没有报错这就尴尬了；查了大量的文档资料；发现这个pool方法都使用了queue.Queue将task传递给工作进程。multiprocessing必须将数据序列化以在进程间传递。方法只有在模块的顶层时才能被序列化，跟类绑定的方法不能被序列化，就会出现上面的异常 ; 那肿么办，我不是一个轻易放弃的人，终于被我找到了方法;

注意
解决方作者是在python3下测试了，python2下用脚本的subprocess要换成value,command = commands.getstatusoutput

解决方法1（亲测）
1.首先要看报错，需要对脚本修改如下:
YaCe类下的multirun方法下修改
for i in range(self.maxpool):
print("开启子进程%s"%i)
p.apply_async(self.conumers,args=(self,i))
为
for i in range(self.maxpool):
print("开启子进程%s"%i)
res = p.apply_async(self.conumers,args=(self,i))
print(res.get)
这就可以看到报错:
cPickle.PicklingError: Can't pickle : attribute lookup builtin.instancemethod failed

2.解决方法如下在脚本中加一个新的函数
(1).def conumers_wrapper(cls_instance,i):
return cls_instance.conumers(i)

(2).修改YaCe下multirun方法
for i in range(self.maxpool):
print("开启子进程%s"%i)
res = p.apply_async(self.conumers,args=(self,i))
print(res.get())
为
for i in range(self.maxpool):
print("开启子进程%s"%i)
res = p.apply_async(conumers_wrapper,args=(self,i))
print(res.get)

问题解决了，运行一下脚本结果还有报错：
RuntimeError: Queue objects should only be shared between processes through inheritance

原因
这里不可以用Queue,要改用Manager.Queue；因为进程之前的同共离用Queue会用问题；

完结
最终代码如下:

==================================================================================
#!/usr/bin/evn python
#_coding:utf8_
from multiprocessing import Pool,Queue,Manager
import time,subprocess,os
class YaCe(object):
def init(self,api,binfile,maxpool,qu,maxrequest=100000,status="success"):
self.api = api
self.binfile = binfile
self.status = status
self.maxpool = maxpool
self.maxrequest = maxrequest
self.qu = qu
def prorequest(self):
for i in range(self.maxrequest):
self.qu.put(i)
print(i)
for i in range(int(self.maxpool)):
self.qu.put(None)
print("None")

def conumers(self,i):
    while True:
        data = self.qu.get(True)
        if data == None:
            print("进程%s任务完成..."%i)
            break
        else:
            #print("time curl -X POST '%s'  --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
            command = subprocess.getoutput("time curl -X POST --connect-timeout 10 '%s'  --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
            #command = subprocess.getoutput("time curl -X POST '%s'  --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
            if self.status == "success":
                logfile = os.getcwd()+"/"+"headbeat.log"+"_%s"%date_time
                if "CgoyMDAwMDAwMDAw" in command:
                    print("进程%s__%s..."%(str(i),str(data)))
                    with open(logfile,"a") as f:
                        f.write(command+"\n")
                    f.close()
                else:
                    with open(logfile,"a") as f:
                        f.write("Faild\n")
                        f.write(command+"\n")
                    f.close()
            else:
                logfile = os.getcwd()+"/"+"roomlist.log"+"_%s"%date_time
                #print("time curl -X POST '%s'  --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
                command = subprocess.getoutput("time curl -X POST --connect-timeout 10 '%s'  --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
                #command = subprocess.getoutput("time curl -X POST '%s' --data-binary @%s"%(self.api,os.getcwd()+"/"+self.binfile))
                if "CAES+" in command:
                    print("进程%s__%s..."%(str(i),str(data)))
                    info = command.split('\n')[-3:]
                    info1 = "\n".join(info)
                    with open(logfile,"a") as f:
                        f.write(info1+"\n")
                    f.close()
                else:
                    print("进程%s__%s..."%(str(i),str(data)))
                    with open(logfile,"a") as f:
                        f.write("Faild\n")
                        f.write(command+"\n")
                    f.close()
def multirun(self):
    ps = int(int(self.maxpool) - 1)
    p = Pool(ps)
    for i in range(self.maxpool):
        print("开启子进程%s"%i)
        p.apply_async(conumers_wrapper,args=(self,i))
    #print(res.get)
    print('等待所有添加的进程运行完毕。。。')
    p.close()
    p.join()
    endtime = time.strftime("%Y%m%d_%X",time.localtime())
    if self.status == "success":
        logfile = os.getcwd()+"/"+"headbeat.log"+"_%s"%date_time
    else:
        logfile = os.getcwd() + "/" + "roomlist.log"+"_%s"%date_time
    with open(logfile,"a") as f:
        f.write("============[%s]============\n"%endtime)
    f.close()
    print('End!!,PID:%s'% os.getpid())

def conumers_wrapper(cls_instance,i):
return cls_instance.conumers(i)

if name == "main":
q = Manager().Queue()
Yc = YaCe('压测接口','二进制证认文件',开多少个进程,queue(队列),maxrequest=100（模拟测试多少次访问）,status="faild"（这里因为测试的两个接口，返回不一样用status参数区分测试的接口的返回值处理）)
Yc.prorequest()
print("++++++")
global date_time
datetime = time.strftime("%Y%m%d%X",time.localtime())
Yc.multirun()

另外有需要云服务器可以了解下创新互联cdcxhl.cn，海内外云服务器15元起步，三天无理由+7*72小时售后在线，公司持有idc许可证，提供“云服务器、裸金属服务器、高防服务器、香港服务器、美国服务器、虚拟主机、免备案服务器”等云主机租用服务以及企业上云的综合解决方案，具有“安全稳定、简单易用、服务可用性高、性价比高”等特点与优势，专为企业上云打造定制，能够满足用户丰富、多元化的应用场景需求。