The present day node-monitoring contains the following:
The collectd custom plugins(majority being sds-specific)
Jinja2 templates for configuring the collectd plugins
A generic flow and a python script(that can be used as a linux command) to get the configs generated from jinja templates given the appropriate parameters.
To be removed:
A python script that pushes collectd detected threshold alert to etcd.
The idea behind a specification for this is to merge node-monitoring and node-agent.
Challenges In consolidating the node-monitoring and node-agent services:
Packaging the sds-specific plugins and templates:
If they are moved out to sds-integrations, then who takes the responsibility of running the generic configuration flow?? Will the flows for even sds configuration still be in node-agent and executed by node-agent itself??
The present day node-monitoring contains the following:
To be removed:
The idea behind a specification for this is to merge node-monitoring and node-agent.
Challenges In consolidating the node-monitoring and node-agent services: