Telegraph precision per input

I am using the telegraf exec input plugin to send some process list to influx and in order to keep the series number low I have a small tag set that initially was overwriting values where there were multiple process of the same name in a single ps listing. I got around that by changing the telegraf agent config to specify microsecond resolution and add 1 us to every point I write in the script that I us to report ps stats.

This is great expect for the fact that I am not using significantly more disk space because precision is global on telegraf and all my input agents have moved from 1s precision to 1us precision. I do not need more than 1s precision expect for this one case.

All the docs and the other posts say that I am SOL in this case as right now there is no way in telegraf to specify precision on a per input basis.

There is one segment of the docs “Precision will NOT be used for service inputs, such as logparser and statsd” which leads me to believe that if I submit to the statsd plugin I can get around this but I was wondering if there is something easier that I can do to get what I want.


We don’t have a way to do this yet, but it would be a nice feature if you would like to open a feature request on github. BTW the service input workaround is pretty clever, you could probably use the socket_listener input listening on a unix socket to do this fairly easily.