python requests 请求报错"cannot
2019-01-11 本文已影响14人
caoxinyiyi
错误描述:
在启动python 多进程跑数据的时候,经常发现真正执行获取数据的子进程莫名其妙的挂掉,在挂掉之前,在记录的info日志中都有一条这样的信息
the error message is cannot concatenate 'str' and 'MaxRetryError' objects
调用的源码
def get(self, url, params={}, headers={}, handle_json=True):
"""requests自动检测编码"""
try:
session = requests.Session()
requests.packages.urllib3.disable_warnings()
response = requests.get(url, params=params, headers=headers, verify=False)
log.info(response)
except Exception, e:
log.info("response error message :" + e.message)
return False
分析原因:
该问题困扰了很久,各种搜索找原因,最终发现的原因是: requests使用了urllib3库,默认的http connection是keep-alive的,http长链接一直不关闭导致的。
解决办法:
将keep_alive的值设置为False
def get(self, url, params={}, headers={}, handle_json=True):
"""requests自动检测编码"""
try:
session = requests.Session()
session.keep_alive = False
requests.packages.urllib3.disable_warnings()
response = requests.get(url, params=params, headers=headers, verify=False)
log.info(response)
except Exception, e:
log.info("response error message :" + e.message)
return False
经过此次调优,报错MaxRetryError 变少。开启了50个进程,有4个进程抛出了这个错误。说明这么解决,还是未能彻底解决问题。根据参考资料,可以将重试次数加大,代码如下
def get(self, url, params={}, headers={}, handle_json=True):
"""requests自动检测编码"""
try:
session = requests.Session()
session.keep_alive = False
requests.adapters.DEFAULT_RETRIES = 5
requests.packages.urllib3.disable_warnings()
response = requests.get(url, params=params, headers=headers, verify=False)
except Exception as e:
log.exception(e)
return False
try:
if handle_json == True:
"requests对于响应的json可以直接获取,并转换成字典"
response = response.json()
except Exception as e:
log.exception(e)
log.info("response data parse to json is error,the error message is " + e.message)
return False