IBM i Prometheus exporter

A configurable Prometheus exporter for IBM i metrics data

Who is this project for?

Companies and organizations with a Grafana + Prometheus stack who want to be able to extract metric data from IBM i such as ASP usage, memory usage, number of jobs running, etc... or even add metrics with custom DB2 queries (using data in QSYS2 or the COUNT() statement).

Installing the exporter on IBM i

To install the exporter, you need to have access to the PASE environment, as well as some open-source software, notably Node.js. Other open-source software is also recommended, such as bash or zsh for a sensible shell, although qsh could do the trick.

Once you have all of that set up, you can move on to these steps:

1. Clone this repo using git, or download it as a zip on a separate computer and transfer it over to the IBM i partition you want to install it on using Putty, scp or FTP.

2. ssh into the machine and cd into the repository

3. Run npm i to install the dependencies

4. Start the exporter with npm start. By default, it runs on TCP port 7203, but you can override this with an environment variable, for example, for a one-off override, you can run it with PORT=8080 npm start

Prometheus configuration

Here is an example prometheus.yml config to scrape data from the exporter:

global:
  scrape_interval: '5s'

scrape_configs:
  - job_name: 'ibmi'
    scrape_interval: '5s'
    static_configs:
      - targets: [ '<IP or hostname of the IBM i partition the exporter is running on>:7203' ]

Adding/removing metrics to query

Adding metrics beyond those already included in metrics.json can require modifying some of the code in src/router.js and src/MetricsQuery.js to fit your needs, however, we think the code is simple enough and you should be able to understand it.

All the metrics that this tool queries are configured in config/metrics.json. The schema this JSON file uses is an array of objects that contains the following:

{
  "name": "Name of the metric",
  "table": "Name of the DB2 table containing these metrics",
  "metricsList": [
    "column 1",
    "column 2",
    "column 3"
  ]
}

Refer to the metrics.json file provided in this repo if you need an example.

The metrics.json file provided by default only includes one DB2 query for some metrics, but it does not include certain metrics such as CPU usage. To add CPU usage metrics, add these lines to config/metrics.json right at the end of the array:

{
  "name": "SYSTEM_ACTIVITY_INFO",
  "table": "QSYS2.SYSTEM_ACTIVITY_INFO",
  "metricsList": [
    "AVERAGE_CPU_RATE",
    "AVERAGE_CPU_UTILIZATION",
    "MINIMUM_CPU_UTILIZATION",
    "MAXIMUM_CPU_UTILIZATION"
  ]
}

Keep in mind that access to QSYS2.SYSTEM_ACTIVITY_INFO requires *JOBCTL authority.