Any plan to support multiple installation? about puppet-telegraf HOT 5 OPEN

Hi, we have the same issue as well. It mostly due to telegraf output all inputs to all outputs which is not always desirable.

Right now we solve it by dumping in another config file, and running a seperate telegraf instance.

from puppet-telegraf.

Why exactly do you need multiple Telegraf instances? As far as I know, you can route the inputs based on a tag to a specific output.

• https://github.com/influxdata/telegraf/blob/master/docs/CONFIGURATION.md#metric-filtering

from puppet-telegraf.

@dhoppe Thanks, I was not aware of that feature.

We do have another case, where we have some system that updates counters every 30 seconds, which we need collect.

And then we have some other collectors on the same hosts that pull data from ~70 system that updates continously. Here we need to add some collection jitter to avoid tipping over the collector VM, and it is not currently possible to have only have collection jitter on some inputs (but I guess that is a telegraf feature request).

from puppet-telegraf.

Telegraf uses an interval of 10s by default, but you can override this for a specific plugin. For example:

[[inputs.snmp]]
  interval = "1m"

Correct. As far as I know, you can not configure a different jitter for a specific plugin.

from puppet-telegraf.

I'm aware of the interval option (we use it widely), but jitter is the problem.

from puppet-telegraf.