Nmon2json

nmon2json - reformat your nmon files to JSON format

nmon2json is a Korn shell script for AIX or Linux to convert nmon collected files to JSON
- which is Python friendly and accepted by online Time-Series database like InfluxDB, Splunk and ELK (logstash)

It is ~600 lines of fairly simple Ksh script, so please fix bugs or add functions and let me know the changes.

Syntax:

nmon2json nmonfile.nmon - outputs to stdout so redirect or pipe some where
nmon2json -h for help & version

Notes for version 27,30 and 31:

Added individual Networks, Disk and Adapter stats
Stats are now split out into sub dictionaries like: "cpu", "lpar", "memory", "disk" etc.
Dates-time strings are now ISO format
Indentation for easy reading
In Python terms, it generates a List of Dictionaries - one Dictionary per snapshot
Version 31 changes to let us use the njmon injector code (with minimal changes) to inject the .json files into InfluxDB

Before you start:

Install InfluxDB
Then create a nmon database within it, like this:

$ influx
> show databases
> create database nmon
> show databases
> exit

Do not insert nmon data into a njmon database = Never tested and more than likely to throw errors as the data format is different.
Also install Grafana - this could be on the InfluxDB computer or locally on your laptop/workstation.

Example:

After the nmon collection is finished

First - convert the raw nmon data (.nmon) file in JSON data (.json)
- nmon2json myfile.nmon >myfile.json
Second - edit the nmon_injector.py and change the host, user and password to match your InfluxDB configuration
Third - inject the JSON (.json) file into InfluxDB
- cat myfile.json | nmon_injector.py

Alternatively, convert and inject in one go into InfluxDB (Assuming you edited the Injector program as above)
- nmon2json myfile.nmon | nmon_injector.py

Example of generated JSON data are in the .tar file for AIX and Linux

Data already supported

Fixes to use ISO standard date-time strings
Stats collection data: hostname, time stamp, Serial number, OS details etc.
Lots of CPU stats
LPAR config and use
Memory, new memory, memory use
Paging
Large page
Proc stats - run queue, system calls etc.
File = directory look-up
Network totals
Network individually - new
Disk totals
Disk individually - new
Adapter individually - new
AIO - Async I/O

Possible "to do" projects:

What would you like next?

JFS?
Network Block Size?
Disk Block Size?
More general config data?
Top Processes?

Sample graphs

nmon to nmon2json to nmon_injector to InfluxDB to Grafana

Downloads

You need BOTH of these files

Download File	Comments
nmon2json_to_InfluxDB_injector.py	21st Feb 2019 - Upload nmon2json output JSON files in to InfluxDB - YOU NEED TO CHANGE THIS FILE BEFORE USE
nmon2json_v32.tar	Version 32 - 9th Dec 2019 - Includes some sample files and the JSON resulting output - v32 handles 68 things on a line like disks. - Above this number it will go horrible wrong. - Contact me if you hit this limit! - Converts nmon files to JSON. - Fixes to match the njmon .JSON output style for an easier life
https://grafana.com/grafana/dashboards?search=nmon2json	Simple Grafana Dashboard for nmon2json graphs. Nothing fancy but might get you started. Hard codes hostname= "engine" in each graph will need changing.