elasticsearch性能测试工具rally深入详解

点此链接

题记

elasticsearch性能测试研究了很久，自己想过通过批量导入数据，然后记录时间，统计CPU、内存等变化，计算得出某个性能指标。但显然，数据量起伏不定，非常不准确。

研究发现，github上提供了rally作为elasticsearch的性能测试工具，较好的解决了es性能测试问题。

1、esrally功能：es的性能测试工具。

esrally不支持windows版本，目前只支持Linux和Mac OS。

esrally的运行基于以下两点假设：

1）所有运行在同一台机器完成。(未来改假设条件可能会去掉）

2）你需要向Elasticsearch索引中添加特定的数据集，然后在其上进行基准查询（benchmarking queries）.

2、esrally安装：

前提:验证是否安装了以下内容（注意版本）

Python 3.4+ available as python3 on the path (verify with: python3 --version which should print Python 3.4.0 or higher)

pip3 available on the path (verify with pip3 --version)

JDK 8

git

如下所示：

[root@laoyang git-2.4.0]# python

Python 3.5.1 (default, Aug 2 2016, 09:53:20)

[GCC 4.4.7 20120313 (Red Hat 4.4.7-17)] on linux

Type "help", "copyright", "credits" or "license" for more information.

[root@laoyang git-2.4.0]# git --version

git version 2.4.0

[root@laoyang git-2.4.0]# java -version

java version "1.8.0_91"

Java(TM) SE Runtime Environment (build 1.8.0_91-b14)

Java HotSpot(TM) 64-Bit Server VM (build 25.91-b14, mixed mode)

[root@laoyang git-2.4.0]# find / -name "pip3"

/usr/bin/pip3

/usr/local/python3/bin/pip3

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

步骤1：安装 Rally

pip3 install esrally

步骤2：配置 Rally

调用： esrally configure 会提示让你输入 java8.home（也就是JAVA_HOME设定的环境变量路径）。

仅java8.home使用自己本机的JAVA_HOME环境变量路径，其他采用默认值即可。

[root@laoyang .rally]# cat /root/.rally/rally.ini

[meta]

config.version = 5

[system]

root.dir = /root/.rally/benchmarks

log.root.dir = logs

env.name = local

[source]

local.src.dir=/root/.rally/src   #注意，此路径并非elaticsearch安装路径。

remote.repo.url=https://github.com/elastic/elasticsearch.git

[provisioning]

local.install.dir = install

[runtime]

java8.home = /opt/jdk1.8.0_91  #注意，JAVA_HOME 路径为：/opt/jdk1.8.0_91

[benchmarks]

local.dataset.cache = ${system:root.dir}/data

[reporting]

datastore.type = elaticsearch

datastore.host = 10.0.1.30

datastore.port = 9200

datastore.secure = False

datastore.user =

datastore.password =

[tracks]

default.url =

https://github.com/elastic/rally-tracks

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

步骤3：运行esrally

由于elasticsearch的运行必须非root账户。esrally建议用非root账户执行。

3、esrally执行常见问题及解决方案

问题1：版本用哪个版本？

目前用最新版本 Rally 0.3.2（2016-8-7发行）。

问题2：注意gradle需要2.13版本，高版本会不兼容。

[root@laoyang caches]# ll /usr/bin/gradle

lrwxrwxrwx 1 root root 27 Aug 5 17:13 /usr/bin/gradle -> /opt/gradle-2.13/bin/gradle

问题3：安装时候提示 documents.json.bz2 无法下载成功。

1）手动下载：

http://benchmarks.elasticsearch.org.s3.amazonaws.com/corpora/geonames/documents.json.bz2

2）移动到 /root/.rally/benchmarks/data/geonames/（对于root用户）

对于非root用户,elaticsearch用户，移动路径为：/home/elasticsearch/benchmarks/data/geonames/

问题4: esrally单独运行和加参数运行区别？

esrally单独运行：非常方便的指令，等价于 :esrally --pipeline=from-sources --version=current. Rally 将要通过Gradle从源码编译elasticsearch。

esrally --pipline=from-distribution --distribution-version=2.3.4, Rally将要从

https://download.elasticsearch.org/elasticsearch/release/org/elasticsearch/distribution/tar/elasticsearch/2.3.4/elasticsearch-2.3.4.tar.gz

下载elaticsearch2.3.4版本运行。

问题5： Could not resolve all dependencies for configuration ‘:benchmarks:classpath’.

37 > Could not download shadow.jar (com.github.jengelman.gradle.plugins:shadow:1.2.3)如何解决？

root账户下：

cd /home/elasticsearch/.rally/benchmarks/src

gradle :distribution:tar:assemble

正确提示信息为：

deprecated API.

Note: Recompile with -Xlint:deprecation for details.

:modules:transport-netty4:processResources UP-TO-DATE

:modules:transport-netty4:classes

:modules:transport-netty4:jar

:modules:transport-netty4:copyPluginPropertiesTemplate

:modules:transport-netty4:pluginProperties

:modules:transport-netty4:bundlePlugin

:distribution:buildModules

:distribution:tar:buildTar

:distribution:tar:generatePomFileForNebulaPublication

:distribution:tar:signArchives SKIPPED

:distribution:tar:assemble

BUILD SUCCESSFUL

4、esrally最终运行结果：

Preparing for race (might take a few moments) ...

 Building from sources ...

/home/elasticsearch/.rally/benchmarks/races/2016-08-08-07-29-59/local/logs/rally_out.logRacing on track [geonames] and challenge [append-no-conflicts] with car [defaults]

 Benchmarking indexing at 12183.1 docs/s [100% done]

 Benchmarking stats (warmup iteration 100/100) [100% done]

 Benchmarking stats (iteration 100/100) [100% done]

 Benchmarking search (warmup iteration 100/100) [100% done]

 Benchmarking search (iteration 100/100) [100% done]

------------------------------------------------------

 _______ __ _____

 / ____(_)___ ____ _/ / / ___/_________ ________

 / /_ / / __ \/ __ `/ / \__ \/ ___/ __ \/ ___/ _ \

/ __/ / / / / / /_/ / / ___/ / /__/ /_/ / / / __/

/_/ /_/_/ /_/\__,_/_/ /____/\___/\____/_/ \___/

| Metric | Value |

|----------------------------------------------------------:|----------:|

| Min Indexing Throughput [docs/s] | 11949 |

| Median Indexing Throughput [docs/s] | 12307 |

| Max Indexing Throughput [docs/s] | 13349 |

| Indexing time [min] | 164.045 |

| Merge time [min] | 32.3815 |

| Refresh time [min] | 8.82333 |

| Flush time [min] | 1.63852 |

| Merge throttle time [min] | 1.45482 |

| Query latency default (90.0 percentile) [ms] | 68.8676 |

| Query latency default (99.0 percentile) [ms] | 77.6009 |

| Query latency default (100 percentile) [ms] | 78.8328 |

| Query latency term (90.0 percentile) [ms] | 4.63227 |

| Query latency term (99.0 percentile) [ms] | 10.9506 |

| Query latency term (100 percentile) [ms] | 13.0573 |

| Query latency phrase (90.0 percentile) [ms] | 6.35018 |

| Query latency phrase (99.0 percentile) [ms] | 13.1745 |

| Query latency phrase (100 percentile) [ms] | 14.2399 |

| Query latency country_agg_uncached (90.0 percentile) [ms] | 361.913 |

| Query latency country_agg_uncached (99.0 percentile) [ms] | 384.75 |

| Query latency country_agg_uncached (100 percentile) [ms] | 392.645 |

| Query latency country_agg_cached (90.0 percentile) [ms] | 7.71451 |

| Query latency country_agg_cached (99.0 percentile) [ms] | 17.4704 |

| Query latency country_agg_cached (100 percentile) [ms] | 18.4851 |

| Query latency scroll (90.0 percentile) [ms] | 50.4234 |

| Query latency scroll (99.0 percentile) [ms] | 50.9866 |

| Query latency scroll (100 percentile) [ms] | 51.2115 |

| Query latency expression (90.0 percentile) [ms] | 482.02 |

| Query latency expression (99.0 percentile) [ms] | 485.951 |

| Query latency expression (100 percentile) [ms] | 491.999 |

| Query latency painless_static (90.0 percentile) [ms] | 890.382 |

| Query latency painless_static (99.0 percentile) [ms] | 918.354 |

| Query latency painless_static (100 percentile) [ms] | 927.282 |

| Query latency painless_dynamic (90.0 percentile) [ms] | 968.908 |

| Query latency painless_dynamic (99.0 percentile) [ms] | 1019.14 |

| Query latency painless_dynamic (100 percentile) [ms] | 1021.97 |

| Median CPU usage (index) [%] | 887.7 |

| Median CPU usage (stats) [%] | 94.9 |

| Median CPU usage (search) [%] | 445.05 |

| Total Young Gen GC [s] | 89.121 |

| Total Old Gen GC [s] | 12.274 |

| Index size [GB] | 3.30111 |

| Totally written [GB] | 20.2123 |

| Heap used for segments [MB] | 21.6794 |

| Heap used for doc values [MB] | 0.113129 |

| Heap used for terms [MB] | 20.0574 |

| Heap used for norms [MB] | 0.0761719 |

| Heap used for points [MB] | 0.770404 |

| Heap used for stored fields [MB] | 0.6623 |

| Segment count | 96 |

| Indices Stats(90.0 percentile) [ms] | 11.5116 |

| Indices Stats(99.0 percentile) [ms] | 14.1724 |

| Indices Stats(100 percentile) [ms] | 36.8348 |

| Nodes Stats(90.0 percentile) [ms] | 11.342 |

| Nodes Stats(99.0 percentile) [ms] | 13.435 |

| Nodes Stats(100 percentile) [ms] | 16.4768 |

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

Logs for this race are archived in /home/elasticsearch/.rally/benchmarks/races/2016-08-08-07-29-59/local/logs-geonames-append-no-conflicts-defaults.zip

5、问题大讨论

https://discuss.elastic.co/t/the-below-bugs-appers-when-running-esrally/57063/25

源码 作者danielmitterdorfer一一回复， 最终作者回复：

I am very happy too and I am glad you had so much persistence. You also uncovered a few usability problems that I need to tackle.

帮作者提了一些建议，促使他升级了一个版本0.3.2。

参考：

https://github.com/elastic/rally

小结

前面的确费劲周折，花费了接近3整天实践，期间甚至告诉原作者工具太难用，给作者反复沟通，最终问题解决。看到输出结果的那一刻，整个人很兴奋。坚持到底，相信付出的力量！不断思考、反思中前行！