too many series for TOP

[aklockmann]

Hi,

Thank you for providing nmon2influxdb. Parsing die pid as a tag in influx results in a unneassary high amount of series cardinality, which is bad for influxdb.
see:
https://www.influxdata.com/blog/red-flags-of-high-cardinality-in-databases/
You cloud just write the pid as an value to influx or remove it.

Kind regards Arne

[adejoux]

Hello,

You are right. I didn't find a good way to store TOP data when it's too big. It's why in most cases, you shoukld exclude this data if you have too many inputs in it.

It's possible to configure nmon to collect less process info.

Regards,

Alain

[toni-moreno]

Hello @adejoux @aklockmann , I would like remember you that there is a parallel project where this issue has been fixed some time ago, and working fine on production today with a more suitable data model for influxdb > 1.0 and with online stats data on data collection, web UI fro configuration and review, and much other improvements if anyone would like to test it.

https://github.com/adejoux/pSeriesCollector

pleased to recover the project if you need @adejoux

[adejoux]

Hello @toni-moreno,

You are completely right. Feel free to recover the project :) I am not working with power systems anymore so it's hard for me to work on this project.

Regards,

Alain