Need clarity for inputs.snmp agents

We are trying to use inputs.snmp to monitor the servers with more than 3000 agents.
So we want to know how many agents can include in the single snmp plugin.
Any limitation to use/include?,
also if we can use more than 3000 agents, do we need to set any other settings for telegraf snmp plugin.

@Hipska you have a large SNMP setup, do you have thoughts on this one?

I would not recommend to specify multiple agents in 1 inputs.snmp config. Although there is no limitation, there are some consequences.

For example if one of those devices is not responding or just very slow, that can take (combined with the timeout and retry settings) more than the set interval to finish. This will result in none of the other devices will get polled the next interval, since the snmp input from previous interval is not completed yet.

This will be eliminated when you create a config section (or even separate file) for each agent. This is easily doable with config management tools like Ansible, Puppet, Salt…

Next to that I would also keep an eye on metric buffer limit, batch sizes and flush intervals. Having that much agents produce a lot of metrics and you don’t want some to be discarded.