我有一个web服务,它接收JSON格式的数据,处理数据,然后将结果返回给请求者。

我想使用cURL测量请求、响应和总时间。

我的请求示例如下:

curl -X POST -d @file server:port

我现在用Linux中的time命令来测量:

time curl -X POST -d @file server:port

然而,time命令只测量总时间——这并不是我想要的。

有什么方法可以使用cURL来测量请求和响应时间吗?


当前回答

以下内容是受西蒙的回答启发。它是自包含的(不需要单独的格式文件),这使得它非常适合包含在.bashrc中。

curl_time() {
    curl -so /dev/null -w "\
   namelookup:  %{time_namelookup}s\n\
      connect:  %{time_connect}s\n\
   appconnect:  %{time_appconnect}s\n\
  pretransfer:  %{time_pretransfer}s\n\
     redirect:  %{time_redirect}s\n\
starttransfer:  %{time_starttransfer}s\n\
-------------------------\n\
        total:  %{time_total}s\n" "$@"
}

此外,它应该适用于curl通常接受的所有参数,因为“$@”只是传递了它们。例如,你可以这样做:

curl_time -X POST -H "Content-Type: application/json" -d '{"key": "val"}' https://postman-echo.com/post

输出:

   namelookup:  0,125000s
      connect:  0,250000s
   appconnect:  0,609000s
  pretransfer:  0,609000s
     redirect:  0,000000s
starttransfer:  0,719000s
-------------------------
        total:  0,719000s

其他回答

您可以使用curl -v——trace-time向跟踪/详细输出添加时间戳。这必须在详细模式或跟踪模式下完成。

测量命令执行时间的通用解决方案

Linux / Mac OS / Unix

在Linux上,您可以使用time命令来测量执行一个命令所花费的时间

time curl https://en.wikipedia.org/wiki/Cat 

输出:


ect...
</body>
</html>

real    0m0.565s
user    0m0.011s
sys     0m0.024s

这里的总时间/响应时间是真实时间。

这个解决方案是通用的,应该适用于任何Unix命令

time sleep 5

输出:

real    0m5.001s
user    0m0.001s
sys 0m0.000s

视窗电源外壳

您可以使用测量命令

如果你想分析或总结延迟,你可以尝试apache bench:

ab -n [number of samples] [url]

例如:

ab -n 100 http://www.google.com/

它将显示:

This is ApacheBench, Version 2.3 <$Revision: 1757674 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking www.google.com (be patient).....done


Server Software:        gws
Server Hostname:        www.google.com
Server Port:            80

Document Path:          /
Document Length:        12419 bytes

Concurrency Level:      1
Time taken for tests:   10.700 seconds
Complete requests:      100
Failed requests:        97
   (Connect: 0, Receive: 0, Length: 97, Exceptions: 0)
Total transferred:      1331107 bytes
HTML transferred:       1268293 bytes
Requests per second:    9.35 [#/sec] (mean)
Time per request:       107.004 [ms] (mean)
Time per request:       107.004 [ms] (mean, across all concurrent requests)
Transfer rate:          121.48 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:       20   22   0.8     22      26
Processing:    59   85 108.7     68     911
Waiting:       59   85 108.7     67     910
Total:         80  107 108.8     90     932

Percentage of the requests served within a certain time (ms)
  50%     90
  66%     91
  75%     93
  80%     95
  90%    105
  95%    111
  98%    773
  99%    932
 100%    932 (longest request)

这是Simons答案的修改版本,使多行输出为单行。它还引入了当前时间戳,以便更容易地跟踪每一行输出。

样本格式

$ cat time-format.txt
time_namelookup:%{time_namelookup} time_connect:%{time_connect} time_appconnect:%{time_appconnect} time_pretransfer:%{time_pretransfer} time_redirect:%{time_redirect} time_starttransfer:%{time_starttransfer} time_total:%{time_total}\n

cmd例子

$ while [ 1 ];do echo -n "$(date) - " ; curl -w @time-format.txt -o /dev/null -s https://myapp.mydom.com/v1/endpt-http; sleep 1; done | grep -v time_total:0

结果

Mon Dec 16 17:51:47 UTC 2019 - time_namelookup:0.004 time_connect:0.015 time_appconnect:0.172 time_pretransfer:0.172 time_redirect:0.000 time_starttransfer:1.666 time_total:1.666
Mon Dec 16 17:51:50 UTC 2019 - time_namelookup:0.004 time_connect:0.015 time_appconnect:0.175 time_pretransfer:0.175 time_redirect:0.000 time_starttransfer:3.794 time_total:3.795
Mon Dec 16 17:51:55 UTC 2019 - time_namelookup:0.004 time_connect:0.017 time_appconnect:0.175 time_pretransfer:0.175 time_redirect:0.000 time_starttransfer:1.971 time_total:1.971
Mon Dec 16 17:51:58 UTC 2019 - time_namelookup:0.004 time_connect:0.014 time_appconnect:0.173 time_pretransfer:0.173 time_redirect:0.000 time_starttransfer:1.161 time_total:1.161
Mon Dec 16 17:52:00 UTC 2019 - time_namelookup:0.004 time_connect:0.015 time_appconnect:0.166 time_pretransfer:0.167 time_redirect:0.000 time_starttransfer:1.434 time_total:1.434
Mon Dec 16 17:52:02 UTC 2019 - time_namelookup:0.004 time_connect:0.015 time_appconnect:0.177 time_pretransfer:0.177 time_redirect:0.000 time_starttransfer:5.119 time_total:5.119
Mon Dec 16 17:52:08 UTC 2019 - time_namelookup:0.004 time_connect:0.014 time_appconnect:0.172 time_pretransfer:0.172 time_redirect:0.000 time_starttransfer:30.185 time_total:30.185
Mon Dec 16 17:52:39 UTC 2019 - time_namelookup:0.004 time_connect:0.014 time_appconnect:0.164 time_pretransfer:0.164 time_redirect:0.000 time_starttransfer:30.175 time_total:30.176
Mon Dec 16 17:54:28 UTC 2019 - time_namelookup:0.004 time_connect:0.015 time_appconnect:3.191 time_pretransfer:3.191 time_redirect:0.000 time_starttransfer:3.212 time_total:3.212
Mon Dec 16 17:56:08 UTC 2019 - time_namelookup:0.004 time_connect:0.015 time_appconnect:1.184 time_pretransfer:1.184 time_redirect:0.000 time_starttransfer:1.215 time_total:1.215
Mon Dec 16 18:00:24 UTC 2019 - time_namelookup:0.004 time_connect:0.015 time_appconnect:0.181 time_pretransfer:0.181 time_redirect:0.000 time_starttransfer:1.267 time_total:1.267

我使用上述方法来捕获上述端点上的慢响应。

Hey比Apache Bench好,有更少的SSL问题

./hey https://google.com -more
Summary:
  Total:    3.0960 secs
  Slowest:  1.6052 secs
  Fastest:  0.4063 secs
  Average:  0.6773 secs
  Requests/sec: 64.5992

Response time histogram:
  0.406 [1] |
  0.526 [142]   |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  0.646 [1] |
  0.766 [6] |∎∎
  0.886 [0] |
  1.006 [0] |
  1.126 [0] |
  1.246 [12]    |∎∎∎
  1.365 [32]    |∎∎∎∎∎∎∎∎∎
  1.485 [5] |∎
  1.605 [1] |

Latency distribution:
  10% in 0.4265 secs
  25% in 0.4505 secs
  50% in 0.4838 secs
  75% in 1.2181 secs
  90% in 1.2869 secs
  95% in 1.3384 secs
  99% in 1.4085 secs

Details (average, fastest, slowest):
  DNS+dialup:    0.1150 secs, 0.0000 secs, 0.4849 secs
  DNS-lookup:    0.0032 secs, 0.0000 secs, 0.0319 secs
  req write:     0.0001 secs, 0.0000 secs, 0.0007 secs
  resp wait:     0.2068 secs, 0.1690 secs, 0.4906 secs
  resp read:     0.0117 secs, 0.0011 secs, 0.2375 secs

Status code distribution:
  [200] 200 responses

参考文献

https://github.com/rakyll/hey