摘要:调大uwsgi配置中 listen=1024的数目是提高并发能力最有效的办法。第二种方法是调大processes数目

发现总是超时报警

1  使用ab确定网络具体征况

[bre@dmp-1 ~]$ ab -n 1000 -c 100 http://.com/Mapping.do?





Concurrency Level:      100

Time taken for tests:   1.040 seconds

Complete requests:      1000

Failed requests:        0

Write errors:           0

Non-2xx responses:      1000

Total transferred:      486000 bytes

HTML transferred:       0 bytes

Requests per second:    961.42 [#/sec] (mean)

Time per request:       104.013 [ms] (mean)

Time per request:       1.040 [ms] (mean, across all concurrent requests)

Transfer rate:          456.30 [Kbytes/sec] received





Connection Times (ms)

              min  mean[+/-sd] median   max

Connect:        0    0   0.4      0       2

Processing:     3   28  61.9     25    1005

Waiting:        3   28  61.9     25    1005

Total:          5   28  61.9     25    1006





Percentage of the requests served within a certain time (ms)

  50%     25

  66%     26

  75%     26

  80%     26

  90%     27

  95%     27

  98%     29

  99%     31

100%   1006 (longest request)

[bre@dmp-1 ~]$ ab -n 1000 -c 200 http://.com/Mapping.do?



Concurrency Level:      200

Time taken for tests:   3.022 seconds

Complete requests:      1000

Failed requests:        0

Write errors:           0

Non-2xx responses:      909

Total transferred:      463432 bytes

HTML transferred:       0 bytes

Requests per second:    330.91 [#/sec] (mean)

Time per request:       604.403 [ms] (mean)

Time per request:       3.022 [ms] (mean, across all concurrent requests)

Transfer rate:          149.76 [Kbytes/sec] received





Connection Times (ms)

              min  mean[+/-sd] median   max

Connect:        0    1   1.2      0       4

Processing:     6  274 791.9     28    3015

Waiting:        6  274 791.9     28    3015

Total:         10  275 792.5     29    3018





Percentage of the requests served within a certain time (ms)

  50%     29

  66%     29

  75%     30

  80%     30

  90%     43

  95%   3015

  98%   3015

  99%   3016

 100%   3018 (longest request)

[bre@dmp-1 ~]$ ab -n 1000 -c 300 http://.com/Mapping.do?





Concurrency Level:      300

Time taken for tests:   3.036 seconds

Complete requests:      1000

Failed requests:        0

Write errors:           0

Non-2xx responses:      811

Total transferred:      439128 bytes

HTML transferred:       0 bytes

Requests per second:    329.35 [#/sec] (mean)

Time per request:       910.892 [ms] (mean)

Time per request:       3.036 [ms] (mean, across all concurrent requests)

Transfer rate:          141.24 [Kbytes/sec] received





Connection Times (ms)

              min  mean[+/-sd] median   max

Connect:        0    2   2.4      0       6

Processing:     5  521 1043.3     29    3027

Waiting:        5  521 1043.3     29    3027

Total:         10  523 1044.8     29    3032





Percentage of the requests served within a certain time (ms)

  50%     29

  66%     31

  75%     35

  80%     53

  90%   3021

  95%   3027

  98%   3029

  99%   3030

 100%   3032 (longest request)

[bre@dmp-1 ~]$ 

[bre@dmp-1 ~]$ ab -n 1000 -c 400 http://.com/Mapping.do?

Concurrency Level:      400

Time taken for tests:   3.025 seconds

Complete requests:      1000

Failed requests:        0

Write errors:           0

Non-2xx responses:      711

Total transferred:      414328 bytes

HTML transferred:       0 bytes

Requests per second:    330.58 [#/sec] (mean)

Time per request:       1209.982 [ms] (mean)

Time per request:       3.025 [ms] (mean, across all concurrent requests)

Transfer rate:          133.76 [Kbytes/sec] received





Connection Times (ms)

              min  mean[+/-sd] median   max

Connect:        0    3   3.7      0       9

Processing:     6  738 1148.8     28    3013

Waiting:        6  738 1148.8     28    3013

Total:         13  741 1151.3     28    3022





Percentage of the requests served within a certain time (ms)

  50%     28

  66%     38

  75%   2011

  80%   2016

  90%   3013

  95%   3016

  98%   3019

  99%   3019

 100%   3022 (longest request)

[bre@dmp-1 ~]$ ab -n 1000 -c 500 http://cm.api.baifendian.com/Mapping.do?bfd_nid=behe

This is ApacheBench, Version 2.3 <$Revision: 655654 $>

Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/

Licensed to The Apache Software Foundation, http://www.apache.org/





Benchmarking cm.api.baifendian.com (be patient)

Completed 100 requests

Completed 200 requests

Completed 300 requests

Completed 400 requests

Completed 500 requests

Completed 600 requests

Completed 700 requests

Completed 800 requests

Completed 900 requests

Completed 1000 requests

Finished 1000 requests









Server Software:        nginx/1.2.9

Server Hostname:        cm.api.baifendian.com

Server Port:            80





Document Path:          /Mapping.do?bfd_nid=behe

Document Length:        0 bytes





Concurrency Level:      500

Time taken for tests:   3.100 seconds

Complete requests:      1000

Failed requests:        0

Write errors:           0

Non-2xx responses:      610

Total transferred:      389280 bytes

HTML transferred:       0 bytes

Requests per second:    322.63 [#/sec] (mean)

Time per request:       1549.754 [ms] (mean)

Time per request:       3.100 [ms] (mean, across all concurrent requests)

Transfer rate:          122.65 [Kbytes/sec] received





Connection Times (ms)

              min  mean[+/-sd] median   max

Connect:        0    5   4.7      9      10

Processing:     9  930 1168.2     31    3014

Waiting:        9  930 1168.2     31    3014

Total:         18  935 1171.1     33    3023





Percentage of the requests served within a certain time (ms)

  50%     33

  66%   2012

  75%   2016

  80%   2018

  90%   3018

  95%   3020

  98%   3021

  99%   3021

 100%   3023 (longest request)

[bre@dmp-1 ~]$ ab -n 1000 -c 600 http://cm.api.baifendian.com/Mapping.do?bfd_nid=behe

This is ApacheBench, Version 2.3 <$Revision: 655654 $>

Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/

Licensed to The Apache Software Foundation, http://www.apache.org/





Benchmarking cm.api.baifendian.com (be patient)

Completed 100 requests

Completed 200 requests

Completed 300 requests

Completed 400 requests

Completed 500 requests

Completed 600 requests

Completed 700 requests

Completed 800 requests

Completed 900 requests

Completed 1000 requests

Finished 1000 requests









Server Software:        nginx/1.2.9

Server Hostname:        cm.api.baifendian.com

Server Port:            80





Document Path:          /Mapping.do?bfd_nid=behe

Document Length:        0 bytes





Concurrency Level:      600

Time taken for tests:   3.039 seconds

Complete requests:      1000

Failed requests:        0

Write errors:           0

Non-2xx responses:      511

Total transferred:      364728 bytes

HTML transferred:       0 bytes

Requests per second:    329.08 [#/sec] (mean)

Time per request:       1823.243 [ms] (mean)

Time per request:       3.039 [ms] (mean, across all concurrent requests)

Transfer rate:          117.21 [Kbytes/sec] received





Connection Times (ms)

              min  mean[+/-sd] median   max

Connect:        0    7   5.2     10      12

Processing:     9 1126 1170.6     45    3022

Waiting:        9 1126 1170.6     45    3022

Total:         21 1133 1173.4     55    3033





Percentage of the requests served within a certain time (ms)

  50%     55

  66%   2014

  75%   2017

  80%   2019

  90%   3019

  95%   3025

  98%   3028

  99%   3031

 100%   3033 (longest request)

可以看到,每秒处理请求数目极限大约是300,随着并发数目的增大,性能开始下降。

现在的请求每天1000000条,那么高峰时段的请求次数为:1000000/(8*60*60)*10 =  347 (次每秒)

每天正常请求时段是8小时,高峰时段请求频次是平均的8~10倍。

所以只是勉强够用而已,还需要提高速度(优化整个程序流程或者增大并发进程)。

2   使用 uwsgitop 查看是前端(nginx)还是后端(uwsgi和后续处理流程)

观察结果

uwsgi-1.9.14 - Mon Feb 10 16:17:16 2014 - req: 188802 - lq: 0 - tx: 66.0M                             node: dmp-1 - cwd: /opt/bre/CookieMapping/dsWebService - uid: 508 - gid: 503 - masterpid: 25062

 WID    %       PID     REQ     EXC     SIG     STATUS  AVG     RSS     VSZ     TX      RunT    

 7      41.4    25073   78127   0       0       idle    1ms     0       0       25.0M   217637

 5      10.0    25071   18956   0       0       idle    1ms     0       0       6.0M    105123

 8      9.1     25076   17213   0       0       idle    1ms     0       0       6.0M    101432

 4      9.1     25070   17196   0       0       idle    1ms     0       0       6.0M    99265

 6      9.0     25072   17032   0       0       idle    1ms     0       0       6.0M    95653

 2      7.6     25068   14312   0       0       idle    1ms     0       0       5.0M    93229

 1      7.1     25067   13489   0       0       idle    17ms    0       0       5.0M    85652

 3      6.6     25069   12477   0       0       idle    2ms     0       0       4.0M    91285

发现当使用ab测试时,当并发比较大时候,时间会出现很大的情况

3  措施

(1)首先,uwsgi观察到上面的负载不均衡,第7个线程负责的请求数远大于其他的。

调整使之均衡

(2)再机器能力足够的情况下,增加线程数目

(3)建立uwsgi集群,在另外一台机器上也部署此服务,并让nginx转发部分到此

uwsgi性能调忧的更多相关文章

  1. centos 7上nginx+uwsgi 性能调优

    上一章将nginx +uwsgi搭建起来,将keystone挂载后面.但是发现一个问题,如果http请求达到一定量后,nginx直接返回502.这让需要部署大规模openstack集群的我很是头疼,比 ...

  2. web前端性能调优

    最近2个月一直在做手机端和电视端开发,开发的过程遇到过各种坑.弄到快元旦了,终于把上线了.2个月干下来满满的的辛苦,没有那么忙了自己准备把前端的性能调优总结以下,以方便以后自己再次使用到的时候得于得心 ...

  3. [网站性能2]Asp.net平台下网站性能调优的实战方案

    文章来源:http://www.cnblogs.com/dingjie08/archive/2009/11/10/1599929.html 前言    最近帮朋友运营的平台进行了性能调优,效果还不错, ...

  4. Asp.net平台下网站性能调优的实战方案(转)

    转载地址:http://www.cnblogs.com/chenkai/archive/2009/11/07/1597795.html 前言 最近帮朋友运营的平台进行了性能调优,效果还不错,所以写出来 ...

  5. 第0/24周 SQL Server 性能调优培训引言

    大家好,这是我在博客园写的第一篇博文,之所以要开这个博客,是我对MS SQL技术学习的一个兴趣记录. 作为计算机专业毕业的人,自己对技术的掌握总是觉得很肤浅,博而不专,到现在我才发现自己的兴趣所在,于 ...

  6. sqlserver性能调优第一步

    相信不少的朋友,无论是做开发.架构的,还是DBA等,都经常听说“调优”这个词.说起“调优”,可能会让很多技术人员心头激情澎湃,也可能会让很多人感觉苦恼,不知道如何入手.当然,也有很多人对此不屑一顾,因 ...

  7. JavaScript:内存泄露、性能调优

    1.在进行JS内存泄露检查之前,先要了解JS的内存管理: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Memory_Manageme ...

  8. hadoop 性能调优与运维

    hadoop 性能调优与运维 . 硬件选择 . 操作系统调优与jvm调优 . hadoop运维 硬件选择 1) hadoop运行环境 2)  原则一: 主节点可靠性要好于从节点 原则二:多路多核,高频 ...

  9. JVM性能调优监控工具jps、jstack、jmap、jhat、jstat、hprof使用详解

    摘要: JDK本身提供了很多方便的JVM性能调优监控工具,除了集成式的VisualVM和jConsole外,还有jps.jstack.jmap.jhat.jstat.hprof等小巧的工具,本博客希望 ...

随机推荐

  1. js连续赋值、指针

    jq的源码中有很多连续赋值,类似这样的: var a = {n:1}; var b = a; // 持有a,以回查 a.x = a = {n:2}; alert(a.x);// --> unde ...

  2. Inno Setup入门(六)——在程序目录下创建文件

    创建文件夹可以使用[dirs]段实现,代码如下: [setup] ;全局设置,本段必须 AppName=Test AppVerName=TEST DefaultDirName="E:\TES ...

  3. rebar

    www.cnblogs.com/panfeng412/archive/2011/08/14/2137990.html

  4. doclint in jdk8

    http://blog.joda.org/2014/02/turning-off-doclint-in-jdk-8-javadoc.html Turning off doclint in JDK 8 ...

  5. drupal错误: Maximum execution time of 240 seconds exceeded

    drupal7.5安装完成,导入汉化包时,出现错误: Fatal error: Maximum execution time of 240 seconds exceeded in D:\phpweb\ ...

  6. vs2010在进行数据架构比较时报'text lines should not be null'错误

    通过VS2010进行服务器数据库和本地数据库比较架构(都是sql server 2008 R2)时,弹出“text lines should be not null”错误,如下图: 解决方法:在Vis ...

  7. php 大转盘抽奖

    包在文件中 lottery.zip <!DOCTYPE HTML><html><head><meta charset="utf-8"> ...

  8. HTML5编程之旅系列一:HTML5 Geolocation 初探

    让我们假设这样一个场景,有一个web应用程序,它可以向用户提供附近不远处某商场的打折优惠信息.使用HTML5 Geolocation API(地理定位 API),可以请求用户共享他们的位置信息. HT ...

  9. CDockablepane风格设置

    屏蔽掉pane右上角的几个按钮 即将CDockablePane右上角的三个按钮屏蔽. 1            去掉关闭按钮 在CDockablePane的派生类中,重写方法CanBeClosed即可 ...

  10. BootStrap详解之(二)

    六.内容 Bootstrap 将全局 font-size 设置为 14px,line-height 设置为 1.428 中心内容 .lead 将字体大小.加粗.行高修改 (无卵高深用) 标记 mark ...