The agent config has grown over time. From my perspective there are 3 main sections in our config:

• outputs: Everything related to configuring where to send data
• datasources: Configurations for the processes the agent runs on how to collect data (soon to be inputs)
• settings: Everything needed to configure the agent itsel

There are few more sections on the top level like management, download, process etc. which this issue is here to discuss.

In summary the suggestion is to move all these to level items under settings. We already have settings.monitoring today. Lets discuss each top level config separately.

Download

Download contains all the settings around from where to fetch binares and where to put them. It is a combination of how to download things and which paths to be used. It is a bit a mixed box and not sure if everything really fits under the same section, but these could go under settings.download for now.

Path (missing?)

There is usage of {path.data} in the config but I couldn't find where these can be configured at the moment?

Process

Process is around how the processes should be run. This could go under settings.process.

Retry

What are these retry configs for? Are these related to fleet / enrollment?

Logging

Logging settings could go under settings.logging.

Management

Management is a mixed box of many things.

There are few parts in here:

• mode: I wonder if we could have fleet.enabled true instead?
• fleet: Do we ever expect users to manually configure this or should the data always com from Fleet for this?
• reporting: This is again a mixed bag. It seems to have configs around logging inside (should this go under logging) and update how to report things to fleet. The Fleet part should go under fleet I assume?
• reload: Is this only related to fleet? If yes, should it go under fleet?

Summary

I plan to update this issue in the end of the discussion with a specific proposal or create a new issue with the proposal we came up with.