ganglia自主监控的思路探讨及实践

最新推荐文章于 2019-08-24 23:47:00 发布

Stephen@Tang

最新推荐文章于 2019-08-24 23:47:00 发布

阅读量1k

点赞数

分类专栏： Ganglia 文章标签： gmetric

Ganglia 专栏收录该内容

3 篇文章 0 订阅

订阅专栏

ganglia是干什么的就不用介绍了，网上很多，我之前的博客里也有详细的讲解。这里主要来讨论一下，ganglia监控本身之外的东西。正刚情况ganglia能够监控这台机器的基本信息，什么cpu,load,memery,swap什么的，都非常详细，但是在实践的工作中，这些数据只能反应出这台机器的基本情况，很难监控到具体业务，所以针对某一些具体业务的监控还是非常重要的。最初我还不是特别明白怎么自己加自己的业务监控，只能在网上找代码，直到我看了mysql的监控之后。

#! /bin/bash

# # Get statistics from MySQL and feed them into Ganglia for monitoring.

# # To use, simply adjust the path to gmetric and mysql credentials as appropriate.

# # Author: David Winterbottom (david.winterbottom@gmail.com)

# Site: http://codeinthehole.com

# Config

# 定义变量 GMETRIC是ganglia的一工具，用它来向ganglia转入数据 其它的看名字应该就知道是干什么的了。

declare -r GMETRIC=/usr/bin/gmetric

declare -r NEW_DATA_FILE=/tmp/mysql-stats.new

declare -r OLD_DATA_FILE=/tmp/mysql-stats.old

declare -r MYSQL_USER="root"

declare -r MYSQL_PASSWORD="insert-password-here"

# Sanity checks  这里是判断gmtric是否存在。正常情况下cluster下都有，但是node机器上可能会没有，这个没关系，直接拷贝一个过去就行了。如果配置文件不对，在对应的文件目录下来一个link -s 就好了

if test -z "$GMETRIC" ; then

printf "The command $GMETRIC is not available";

exit 192

fi

# Function for submiting metrics 通过metric向ganglia提供数据。其实我要说的精髓就是这里。gmetric --type 类型 --name 名字 --value 值 --unit 更新频率

function record_value {

if [ $# -lt 1 ]; then

printf "You must specify a look-up value\n"

exit 192

fi

LOOKUP_VAR=$1

GANGLIA_NAME=${2-unspecified}

GANGLIA_TYPE=${3-float}

GANGLIA_UNITS=${4-units}

GANGLIA_VALUE=`grep "$LOOKUP_VAR[^_]" "$NEW_DATA_FILE" | awk '{print $2}'` printf " * $GANGLIA_NAME: $GANGLIA_VALUE\n"

$GMETRIC --type "$GANGLIA_TYPE" --name "$GANGLIA_NAME" --value $GANGLIA_VALUE --unit "$GANGLIA_UNITS" }

# Function for submitting delta metrics 这个就是文件中收集数据并向ganglia发送数据了。

function record_value_rate

{

   if [ $# -lt 1 ]; then

       printf "You must specify a look-up value\n"

       exit 192

   fi

   MYSQL_VAR=$1

   GANGLIA_NAME=${2-unspecified}

   GANGLIA_TYPE=${3-float}

   GANGLIA_UNITS=${4-"per second"}

   # Get values from old and new files

   PREVIOUS_VALUE=`grep "$MYSQL_VAR[^_]" "$OLD_DATA_FILE" | awk '{print $2}'`

   NEW_VALUE=`grep "$MYSQL_VAR[^_]" "$NEW_DATA_FILE" | awk '{print $2}'`

   DELTA_VALUE=$[ $NEW_VALUE-$PREVIOUS_VALUE ]

   PREVIOUS_TIMESTAMP=`date -r "$OLD_DATA_FILE" +%s`

   NEW_TIMESTAMP=`date -r "$NEW_DATA_FILE" +%s`

   DELTA_TIMESTAMP=$[ $NEW_TIMESTAMP-$PREVIOUS_TIMESTAMP ]

   if [ $DELTA_VALUE -lt 0 ] || [ $DELTA_TIMESTAMP -lt 0 ]; then

       # Something strange here - MYSQL may just have started.  Ignore for now

       printf "Weird data value - skipping\n"

   else

       # Need to pipe to bc to perform floating point operations

       DELTA_RATE=`echo "scale=4; $DELTA_VALUE/$DELTA_TIMESTAMP" | bc -l`

       printf " * $GANGLIA_NAME -- Previous value: $PREVIOUS_VALUE, new value: $NEW_VALUE, delta: $DELTA_VALUE, previous timestamp: $PREVIOUS_TIMESTAMP, new timestamp: $NEW_TIMESTAMP, delta: $DELTA_TIMESTAMP, $DELTA_RATE per second\n"

       $GMETRIC --type "$GANGLIA_TYPE" --name "$GANGLIA_NAME" --value $DELTA_RATE --unit "$GANGLIA_UNITS"

   fi

}

# Read MySQL statistics into a temporary file 这个基本都能看懂了，就是去mysql查看信息

mysql --user=$MYSQL_USER --password=$MYSQL_PASSWORD --execute "SHOW GLOBAL STATUS" > "$NEW_DATA_FILE"

# Submit metrics 监控mysql的数据

record_value_rate "Connections" "MYSQL_CONNECTIONS" "float" "Connections/sec"

record_value_rate "Com_update" "MYSQL_UPDATE_QUERIES" "float" "Queries/sec"

record_value_rate "Com_select" "MYSQL_SELECT_QUERIES" "float" "Queries/sec"

record_value_rate "Com_insert" "MYSQL_INSERT_QUERIES" "float" "Queries/sec"

record_value_rate "Com_delete" "MYSQL_DELETE_QUERIES" "float" "Queries/sec"

record_value_rate "Created_tmp_tables" "MYSQL_CREATED_TMP_TABLES" "float" "Tables created/sec"

record_value_rate "Slow_queries" "MYSQL_SLOW_QUERIES" "float" "Queries/sec"

record_value_rate "Qcache_hits" "MYSQL_QUERY_CACHE_HITS" "float" "Hits/sec"

record_value "Qcache_queries_in_cache" "MYSQL_QUERIES_IN_CACHE" "float" "Queries"

record_value_rate "Questions" "MYSQL_QUESTIONS" "float" "Questions/sec"

record_value_rate "Threads_connected" "MYSQL_THREADS_CONNECTED" "float" "Threads connected/sec"

record_value "Threads_running" "MYSQL_THREADS_RUNNING" "float" "Threads running"

# Copy data 覆盖旧数据。

cp "$NEW_DATA_FILE" "$OLD_DATA_FILE"

代码分析完了，我相信只要稍微懂点shell和mysql的人都能看懂。而通过这个脚本，我学到了什么呢？1，复制，用这套代码稍微修改一下是不是可以监控redis，tomcat什么的呢？ 2，关键点，其实就一个gmetric。把这个的参数弄懂完就差不多了。我想了一下，参数是不是只有四个呢？本着寻根究底的精神，运行了一下 gmetric --help （大多数情况下，这个都是很有用的）于是出现以下结果

-h, --help Print help and exit

-V, --version Print version and exit

-c, --conf=STRING The configuration file to use for finding send channels

(default=`/usr/local/etc/gmond.conf')

-n, --name=STRING Name of the metric

-v, --value=STRING Value of the metric

-t, --type=STRING Either

-u, --units=STRING Unit of measure for the value e.g. Kilobytes, Celcius

(default=`')

-s, --slope=STRING Either zero|positive|negative|both (default=`both')

-x, --tmax=INT The maximum time in seconds between gmetric calls

(default=`60')