jmx_exporter文档https://github.com/prometheus/jmx_exporter/blob/master/README.md

原生的jmx_exporter需要为每一个java进程启动一个exporter，并且需要部署到该进程所属的机器上。可能存在一台机器上有很多jvm进程，那么一台机器上也需要部署多个exporter。为了方便管理，我希望能够在一台机器上部署一个jmx_exporter.

这里提供了一种解决方案：将一个jmx_exporter部署在一台机器上，同时抓取多个应用的jmx。现对于原生的jmx_exporter,只需要修改很少的配置文件即可。

比如抓取两个进程的jmx，只需要将原生配置改为列表即可，增加了几个变量jobs,name,socket

删除了javaagent启动方式

确保被抓取目标开启了remote jmx，并且配置文件无误

java -jar jmx_prometheus_httpserver_multi-0.13.0-SNAPSHOT-jar-with-dependencies.jar 8 example.yaml

参数分别是：你要抓取的jvm进程数，配置文件

The configuration is in YAML. An example with all possible options:

---

jobs:
  - name: a
    startDelaySeconds: 0
    #hostPort: 127.0.0.1:1234
    username: 
    password: 
    socket: 127.0.0.1:9999
    jmxUrl: service:jmx:rmi:///jndi/rmi://127.0.0.1:1234/jmxrmi
    ssl: false
    lowercaseOutputName: false
    lowercaseOutputLabelNames: false
    whitelistObjectNames: ["org.apache.cassandra.metrics:*"]
    blacklistObjectNames: ["org.apache.cassandra.metrics:type=ColumnFamily,*"]
    rules:
    - pattern: 'org.apache.cassandra.metrics<type=(\w+), name=(\w+)><>Value: (\d+)'
    name: cassandra_$1_$2
    value: $3
    valueFactor: 0.001
    labels: {}
    help: "Cassandra metric $1 $2"
    type: GAUGE
    attrNameSnakeCase: false
	
 - name: a
   startDelaySeconds: 0
   hostPort: 127.0.0.1:1235
   username: 
   password: 
   socket: 127.0.0.1:9999
   #jmxUrl: service:jmx:rmi:///jndi/rmi://127.0.0.1:1235/jmxrmi
   ssl: false
   lowercaseOutputName: false
   lowercaseOutputLabelNames: false
   whitelistObjectNames: ["org.apache.cassandra.metrics:*"]
   blacklistObjectNames: ["org.apache.cassandra.metrics:type=ColumnFamily,*"]
   rules:
   - pattern: 'org.apache.cassandra.metrics<type=(\w+), name=(\w+)><>Value: (\d+)'
   name: cassandra_$4_$5
   value: $5
   valueFactor: 0.001
   labels: {}
   help: "Cassandra metric $4 $5"
   type: GAUGE
   attrNameSnakeCase: false

Metric names and label names are sanitized. All characters other than [a-zA-Z0-9:_] are replaced with underscores, and adjacent underscores are collapsed. There's no limitations on label values or the help text.

A minimal config is {}, which will connect to the local JVM and collect everything in the default format. Note that the scraper always processes all mBeans, even if they're not exported.

Example configurations for javaagents can be found at https://github.com/prometheus/jmx_exporter/tree/master/example_configs

The format of the input matches against the pattern is

domain<beanpropertyName1=beanPropertyValue1, beanpropertyName2=beanPropertyValue2, ...><key1, key2, ...>attrName: value

No escaping or other changes are made to these values, with the exception of if attrNameSnakeCase is set. The default help includes this string, except for the value.

The default format will transform beans in a way that should produce sane metrics in most cases. It is

domain_beanPropertyValue1_key1_key2_...keyN_attrName{beanpropertyName2="beanPropertyValue2", ...}: value

If a given part isn't set, it'll be excluded.

You can start the jmx's scraper in standalone mode in order to debug what is called

java -cp jmx_exporter.jar io.prometheus.jmx.JmxScraper service:jmx:rmi:your_url

To get finer logs (including the duration of each jmx call), create a file called logging.properties with this content:

handlers=java.util.logging.ConsoleHandler
java.util.logging.ConsoleHandler.level=ALL
io.prometheus.jmx.level=ALL
io.prometheus.jmx.shaded.io.prometheus.jmx.level=ALL

Add the following flag to your Java invocation:

-Djava.util.logging.config.file=/path/to/logging.properties