A new tool for generating Graphite metrics from PostgreSQL queries

[Gro-Anita Vindheim]

Because comments can't be edited, I will post the implementation notes above the original request.

This will be implemented as a NAV subsystem. This means that it is
started with the NAV command. The suggested name is 'stats'. The job
itself will run as a cron job.

The code itself will use the 'send_metrics' function to send collected
data to Carbon. This ensures correctness.

To make collection flexible, the collection parameters - what to
collect and where to send it - will be put in a configuration file
called 'stats.conf'. Each collection have a section. Each section has
two required options - the metric path and the SQL to run to get the
metric value to send. The metric timestamp will be fetched when the
metric is sent.

A new metric path prefix is created for these data - nav.stats.

The initial proposal is kinda light on specs but suggested paths for
the wishes are:
- "How many switchports, and how many of those are in use"
  - nav.stats.ports.switchports.active
  - nav.stats.ports.switchports.all
- "What kind of netboxes do we have, sort by vendor & model"
  - nav.stats.devices.<vendor>.<type>
- "how many netboxes do we have in our network" - sum of the above

Original request follows:

For the wishlist...

Our management often ask for information like
"how many netboxes do we have in our network"
"How many switchports, and how many of those are in use"
"What kind of netboxes do we have, sort by vendor & model"

I would like NAV to save these data as carbon-datapoints.

This will make my management able too see trends without asking me once a month :)

[John-Magne Bredal (john-m-bredal)]

I guess this is relevant for the new WatchDog overview, as there already are some NAV overview data there.

Personally I think it's a good idea to store this data using Carbon.