websocket长连接压力测试踩过的坑
Websocket协议压测记录
背景:
公司的行情系统是采用的websocket协议,有请求和订阅两种方式向服务器申请最新行情信息。请求方式是一次的,订阅方式是建立连接后,服务器定时向客户端推送行情信息。
初步测试方案:
因考虑到websocket是双工通讯,是长连接,并且本次压测的性能指标是系统能建立的最大连接数,并且是建立连接后服务器能持续向客户端推送行情信息。
基于以上原因考虑用python采用多线程建立连接,为了验证能否收到推送的信息,把返回的行情信息保存到文本文件中。Python脚本如下:
import websocket
import time
import threading
import gzip
#import json
#from threadpool import ThreadPool, makeRequests
#from websocket import create_connection
SERVER_URL = "ws://pro-web-new.devtest.exshell-dev.com/r1/main/ws"
#SERVER_URL = "wss://i.cg.net/wi/ws"
#SERVER_URL = "wss://www.exshell.com/r1/main/ws"
def on_message(ws, message):
print(message)
def on_error(ws, error):
print(error)
def on_close(ws):
print("### closed ###")
def on_open(ws):
def send_trhead():
send_info = '{"sub": "market.ethusdt.kline.1min","id": "id10"}'
#send_info = '{"event":"subscribe", "channel":"btc_usdt.deep"}'
while True:
#time.sleep(5)
#ws.send(json.dumps(send_info))
ws.send(send_info)
while (1):
compressData = ws.recv()
result = gzip.decompress(compressData).decode('utf-8')
if result[:7] == '{"ping"':
ts = result[8:21]
pong = '{"pong":' + ts + '}'
ws.send(pong)
ws.send(send_info)
else:
#print(result)
with open('./test_result.txt', 'a') as f:
f.write(threading.currentThread().name+'\n')
f.write(result+'\n')
t = threading.Thread(target=send_trhead)
t.start()
print(threading.currentThread().name)
def on_start(a):
# time.sleep(2)
# websocket.enableTrace(True)
# ws = websocket.WebSocketApp(SERVER_URL,
# on_message=on_message,
# on_error=on_error,
# on_close=on_close)
# ws.on_open = on_open
# ws.run_forever()
#print(a[2])
try:
ws = websocket.create_connection(SERVER_URL)
on_open(ws)
except Exception as e:
print('error is :',e)
print('connect ws error,retry...')
time.sleep(5)
if __name__ == "__main__":
# pool = ThreadPool(3)
# test = list()
# for ir in range(3):
# test.append(ir)
#
# requests = makeRequests(on_start, test)
# [pool.putRequest(req) for req in requests]
# pool.wait()
# # #on_start(1)
for ir in range(20):
on_start(1)
time.sleep(0.1)
初步测试结果:
在压测的过程中,发现连接数达到一定程度(单机1400连接),连接就断掉了,监控发现压力机内存基本消耗光了,因建立连接,并接收返回的信息,随着连接数增加,内存消耗大,只能断开连接,释放内存。
调整测试方案:
和架构、开发讨论后,准备在websocket客户端采用AIO异步通讯方式增大压力,因当时是考虑到长连接未考虑这种方式,查询资料,发现websocket服务端可以采用AIO异步通讯方式,在websocket客户端尝试一下,采用locust + python的方式,也查找了一些资料,发现方案可行。
Locust是一款可扩展的,分布式的,性能测试的,开源的,用Python编写的性能测试工具。对于测试HTTP协议的接口是比较方便的,但是它也支持测试别的协议的接口,不过需要重写Locust类。脚本如下:
from locust import Locust, events, task, TaskSet
import websocket
import time
import gzip
class WebSocketClient():
def __init__(self, host, port):
self.host = host
self.port = port
class WebSocketLocust(Locust):
def __init__(self, *args, **kwargs):
self.client = WebSocketClient("172.31.15.85", 9503)
class UserBehavior(TaskSet):
ws = websocket.WebSocket()
#self.ws.connect("ws://10.98.64.103:8807")
ws.connect("ws://pro-web-new.devtest.exshell-dev.com/r1/main/ws")
@task(1)
def buy(self):
try:
start_time = time.time()
#self.ws.send('{"url":"/buy","data":{"id":"123","issue":"20170822","doubled_num":2}}')
#result = self.ws.recv()
send_info = '{"sub": "market.ethusdt.kline.1min","id": "id10"}'
# send_info = '{"event":"subscribe", "channel":"btc_usdt.deep"}'
while True:
# time.sleep(5)
# ws.send(json.dumps(send_info))
ws.send(send_info)
while (1):
compressData = ws.recv()
result = gzip.decompress(compressData).decode('utf-8')
if result[:7] == '{"ping"':
ts = result[8:21]
pong = '{"pong":' + ts + '}'
ws.send(pong)
ws.send(send_info)
else:
# print(result)
with open('./test_result.txt', 'a') as f:
#f.write(threading.currentThread().name + '\n')
f.write(result + '\n')
except Exception as e:
print("error is:",e)
class ApiUser(WebSocketLocust):
task_set = UserBehavior
min_wait = 100
max_wait = 200
用命令执行脚本:
Locust -f websocket_client_locust.py –-no-web -c 1 -r 1 -t 60s
单个用户执行成功,并能生成文件。但多个用户执行的时候就报错,报错信息如下:This socket is already used by another greenlet: <bound method Waiter.switch of <gevent.hub.Waiter object at 0x7f01f0594900>>
错误原因说的是socke正在被使用,但是我的代码中是新的socket,简单分析了一下,应该不会出现问题,但是我的socek的使用部分是一个全局的client,然后程序运行的时候出现了上述错误.仔细推测我找出了原因:
geven是个协程库,那么多个协程共用一个socek的时候就会出现上述错误了,于是我把socket改成了局部的,问题解决.
修改前:
修改后:
修改后代码:
from locust import Locust, events, task, TaskSet
import websocket
import time
import gzip
class WebSocketClient():
def __init__(self, host):
self.host = host
#self.port = port
class WebSocketLocust(Locust):
def __init__(self, *args, **kwargs):
self.client = WebSocketClient("172.31.15.85")
class UserBehavior(TaskSet):
# ws = websocket.WebSocket()
# #self.ws.connect("ws://10.98.64.103:8807")
# ws.connect("ws://pro-web-new.devtest.exshell-dev.com/r1/main/ws")
@task(1)
def buy(self):
try:
ws = websocket.WebSocket()
# self.ws.connect("ws://10.98.64.103:8807")
ws.connect("ws://pro-web-new.devtest.exshell-dev.com/r1/main/ws")
start_time = time.time()
#self.ws.send('{"url":"/buy","data":{"id":"123","issue":"20170822","doubled_num":2}}')
#result = self.ws.recv()
send_info = '{"sub": "market.ethusdt.kline.1min","id": "id10"}'
# send_info = '{"event":"subscribe", "channel":"btc_usdt.deep"}'
while True:
# time.sleep(5)
# ws.send(json.dumps(send_info))
ws.send(send_info)
while (1):
compressData = ws.recv()
result = gzip.decompress(compressData).decode('utf-8')
if result[:7] == '{"ping"':
ts = result[8:21]
pong = '{"pong":' + ts + '}'
ws.send(pong)
ws.send(send_info)
# else:
# # print(result)
# with open('./test_result.txt', 'a') as f:
# #f.write(threading.currentThread().name + '\n')
# f.write(result + '\n')
except Exception as e:
print("error is:",e)
class ApiUser(WebSocketLocust):
task_set = UserBehavior
min_wait = 100
max_wait = 200
压测开始,随着用户数上升,压力机端发生如下错误:500和502错误
这是协议进行握手时失败,查询后端行情应用服务器,也有大量报错。
查看服务器发现打开最大文件数是1024,调整到65535,用如下命令调整:
第一步,修改/etc/sysctl.conf文件,在文件中添加如下行:
net.ipv4.ip_local_port_range = 1024 65000
这表明将系统对本地端口范围限制设置为1024~65000之间。请注意,本地端口范围的最小值必须大于或等于1024;而端口范围的最大值则应小于或等于65535.修改完后保存此文件。
第二步,执行sysctl命令:
[speng@as4 ~]$ sysctl -p
如果系统没有错误提示,就表明新的本地端口范围设置成功。如果按上述端口范围进行设置,则理论上单独一个进程最多可以同时建立60000多个TCP客户端连接。
调整完成后复测,发现还是报这个错误,请开发进行定位分析应用程序。