2023年12月6日发(作者：)
DNS服务部署的那点事儿之DNS压力测试
#刚才发博文提示大于15万字儿，写不开了，单独开篇博文来补充一下
Bind压力测试
bind有专门的工具可以将配置文件的数据放入到数据库中，对于bind来讲，最常见的压力测试软件是querypery，此工具是bind源码包自带的
一个工具，但安装完bind之后很有可能官方的包并没有给我们提供由此我们不得不去自己编译安装它
1.安装queryperf
可以在 下载bind10-1.1.0并安装它们
[root@testtools]# cd bind-9.9.4-P1/contrib/queryperf/
[root@testqueryperf]# make
gcc -DHAVE_CONFIG_H -c queryperf.c
gcc -DHAVE_CONFIG_H queryperf.o -lnsl -lresolv -lm -o queryperf
#不用直接make install 将queryperf 拷贝至/bin/目录即可
[root@testqueryperf]# cp queryperf /bin/
2.使用queryperf
首先创建文件，指定需要测试的域名和记录类型：
[root@test~]# cat
10.0.10.60PTR
使用-d指定数据文件，做压力测试的时候，要对其解析哪些记录为基准型测试，-s 指定需要压测的服务器IP
[root@test~]# queryperf -d -s 10.0.10.60
DNSQuery Performance Testing Tool
Version:$Id: queryperf.c,v 1.12 2007/09/05 07:36:04 marka Exp $
[Status]Processing input data #正在读取数据
[Status]Sending queries (beginning with 10.0.10.60) #发送查询请求
[Status]Testing complete
Statistics:
Parse input file: once
Ended due to: reaching end of file Queries sent: 3 queries #共发起了多少请求
Queries completed: 3 queries #完成了多少请求
Queries lost: 0 queries #丢失多少请求
Queries delayed(?): 0 queries
RTT max: 0.006557sec #将请求发出去响应回过来最大的一次耗时多少
RTT min: 0.002860 sec #最短一次
RTT average: 0.004116 sec #平均多久
RTT std deviation: 0.001721 sec
RTT out of range: 0 queries
Percentage completed: 100.00% #完成的比例
Percentage lost: 0.00% #丢失的比例
Started at: Wed Dec 18 20:29:43 2013
Finished at: Wed Dec 18 20:29:43 2013
Ran for: 0.009473 seconds
Queries per second: 316.689539 qps #每秒钟平均能完成多少个请求
#之前测试的数量太少，接下来修改,将信息填满：
10.0.10.60PTR
10.0.10.61PTR
10.0.10.60PTR
10.0.10.61PTR
10.0.10.60PTR
#这个域名不存在
然后将如上行多复制几行，（不论怎么复制行都可以）如下所示：
[root@test~]# wc -l
再次进行压力测试[root@test~]# queryperf -d -s 10.0.10.60DNSQuery Performance Testing ToolVersion:$Id: queryperf.c,v 1.12 2007/09/05 07:36:04 marka Exp $[Status]Processing input data[Status]Sending queries (beginning with 10.0.10.60)[Timeout]Query timed out: msg id 62[Timeout]Query timed out: msg id 64[Timeout]Query timed out: msg id 71[Timeout]Query timed out: msg id 73[Timeout]Query timed out: msg id 80[Timeout]Query timed out: msg id 82[Timeout]Query timed out: msg id 89[Timeout]Query timed out: msg id 91[Timeout]Query timed out: msg id 98[Timeout]Query timed out: msg id 100[Status]Testing completeStatistics: Parse input file: once Ended due to: reaching end of file Queries sent: 2364 queries Queries completed: 2364 queries Queries lost: 0 queries Queries delayed(?): 0 queries RTT max: 0.347522sec RTT min: 0.000083 sec RTT average: 0.002009 sec RTT std deviation: 0.022389 sec RTT out of range: 0 queries Percentage completed: 100.00%
Percentage lost: 0.00%
Started at: Wed Dec 18 20:40:35 2013
Finished at: Wed Dec 18 20:40:40 2013
Ran for: 5.024620 seconds
Queries per second: 470.483340 qps
#可以看到以下有很多包处于丢失状态
[Status]Processing input data
[Status]Sending queries (beginning with 10.0.10.60)
[Timeout]Query timed out: msg id 62
[Timeout]Query timed out: msg id 64
[Timeout]Query timed out: msg id 71
[Timeout]Query timed out: msg id 73
[Timeout]Query timed out: msg id 80
#----------略------------------
[Status]Testing complete
但是在本地测试不用考虑服务器端的带宽，查看本地服务器负载情况：
[root@test~]# uptime
20:51:07 up 2:27, 2 users, load average: 2.20, 0.00, 0.00
发现利用率也不高，使用vmstat查看实时负载情况，可以看到，阻塞队列正常，bi 和 bo量偶尔会有所以中断和上下文切换量也很小，所以
这些都不是问题，可是响应速度依然非常的慢
[root@test~]# vmstat 1
procs-----------memory---------- ---swap-- -----io---- --system-- -----cpu-----
r b swpd free buff cache si so bi bo in cs us sy id wa st
0 0 0 68720 53988 214796 0 0 26 15 25 23 0 0 96 3 0
0 0 0 68712 53988 214800 0 0 0 0 18 20 0 0 100 0 0
0 0 0 68712 53988 214800 0 0 0 0 19 13 0 0 100 0 0
0 0 0 68712 53988 214800 0 0 0 0 15 14 0 1 99 0 0
技术压力测试，切换到其它服务器对其进行压力测试
[root@test~]# queryperf -d -s 10.0.10.61
依旧如此有一堆的timeoutWarning:RTT is out of range: 29.990839 [query=10.0.10.61/12, rcode=2]
Warning:RTT is out of range: 24.962086 [query=10.0.10.61/12, rcode=2]
Warning:RTT is out of range: 24.962383 [query=10.0.10.61/12, rcode=2]
[Timeout]Query timed out: msg id 223
[Timeout]Query timed out: msg id 224
[Timeout]Query timed out: msg id 212
[Timeout]Query timed out: msg id 193
接下来在被测试的服务器上使用rndcstatus查看named服务器的运行状况
[root@test2~]# rndc status
version:6_4.6
CPUsfound: 1
workerthreads: 1
numberof zones: 21
debuglevel: 0
xfersrunning: 0
xfersdeferred: 0
soaqueries in progress: 0
querylogging is OFF #query log是关闭的
recursiveclients: 20/0/1000 #收到20个递归请求 tcp clien没有
tcpclients: 0/100
serveris up and running
怀疑是否是反向解析的问题，将包含反向解析的行全部使用sed删除
[root@test~]# sed -i '/^10/d'
[root@test~]# queryperf -d -s 10.0.10.61
DNSQuery Performance Testing Tool
Version:$Id: queryperf.c,v 1.12 2007/09/05 07:36:04 marka Exp $
[Status]Processing input data
[Status]Sending queries (beginning with 10.0.10.61)
[Timeout]Query timed out: msg id 1
[Timeout]Query timed out: msg id 2
[Timeout]Query timed out: msg id 3
[Timeout]Query timed out: msg id 4[Timeout]Query timed out: msg id 5
[Timeout]Query timed out: msg id 6
[Timeout]Query timed out: msg id 7
[Timeout]Query timed out: msg id 8
[Timeout]Query timed out: msg id 9
Statistics:
Parse input file: once
Ended due to: reaching end of file
Queries sent: 4548 queries
Queries completed: 4548 queries
Queries lost: 0 queries
Queries delayed(?): 0queries
RTT max: 0.004176sec
RTT min: 0.000126 sec
RTT average: 0.001663 sec
RTT std deviation: 0.000515 sec
RTT out of range: 0 queries
Percentage completed: 100.00%
Percentage lost: 0.00%
Started at: Wed Dec 18 20:59:48 2013
Finished at: Wed Dec 18 20:59:48 2013
Ran for: 0.405095 seconds
Queries per second: 11226.996137 qps #总共每秒可以承受11226个查询请求
进一步压力测试：
可以看到，此服务器负载比刚才高出了一些
[root@test~]# wc -l
[root@test~]# queryperf -d -s 10.0.10.61[root@test2~]# vmstat 1procs-----------memory---------- ---swap-- -----io---- --system-- -----cpu-----r b swpd free buff cache si so bi bo in cs us sy id wa st1 0 0 314332 12800 60260 0 0 2 0 7686 59 25 95 0 0 0
1 0 0 314332 12800 60260 0 0 0 0 7731 36 30 89 1 0 0
1 0 0 314332 12800 60260 0 0 0 0 7494 40 36 84 0 0 0得出结果：DNSQuery Performance Testing ToolVersion:$Id: queryperf.c,v 1.12 2007/09/05 07:36:04 marka Exp $[Status]Processing input data[Status]Sending queries (beginning with 10.0.10.61)[Status]Testing completeStatistics: Parse input file: once Ended due to: reaching end of file Queries sent: 3774840 queries Queries completed: 3774840 queries Queries lost: 0 queries Queries delayed(?): 0 queries RTT max: 0.041022sec RTT min: 0.000037 sec RTT average: 0.001813 sec RTT std deviation: 0.000452 sec RTT out of range: 0 queries Percentage completed: 100.00% Percentage lost: 0.00% Started at: Wed Dec 18 21:04:24 2013
Finished at: Wed Dec 18 21:10:16 2013
Ran for: 351.908312 seconds
Queries per second: 8726.771353 qps
感兴趣的童鞋，可以将查询日志打开，再次进行压力测试，这里就不做演示了，打开后可以通过iostat 来查看负载情况，每秒读写的数据量
多少，很有可能在某个磁盘上不停的写入数据，所以正是因为io子系统比较慢
3.使用dnstop监视bind服务性能状态
使用dnstop来实现观测dns响应请求查询时候的性能
安装dnstop
[root@testdnstop-20121017]# cd dnstop-20121017
[root@testdnstop-20121017]# ./configure
[root@testdnstop-20121017]# make && makeinstall
开始抓包分析，对eth0接口的请求进行分析：
[root@testdnstop-20121017]# dnstop -4 -Q -R eth0
Queries:0 new, 0 total Wed Dec 18 21:59:57 2013
Replies:0 new, 0 total
Sources Count % cum%
到10.0.10.61上对其做dig查询
[root@test2tools]# dig -t A @10.0.10.60
[root@test2tools]# dig -t A @10.0.10.60
再次查看60的监测状态：
Queries:0 new, 2 total
Wed Dec18 22:01:01 2013
Replies:0 new, 2 total
Sources Count % cum%
------------------- ------ ------
10.0.10.61 2 100.0 100.0
可以键1-3对其对类型进行统计比如
com
总结：
（1）抓包时可以在61上对60进行压力测试观测效果）；
（2）建议做压力测试的时候对目标主机做分布式压力测试，效果更佳；
本文转自zuzhou 51CTO博客，原文链接:/yijiu/1343257
发布评论