Reporting metrics to Graphite/StatsD

I'm currently thinking how I can report metrics to our graphite or statsd instance.

What would be a better solution? I could have a periodical push that sends the current counts of active, failed and completed jobs or I could send an increment on each job change.

What I like about the periodical push is, that I would definitely be notified about how many jobs are currently stuck. I could probably calculate that with a difference between the active and completed and failed jobs when I would increment them.

What solutions have you come up with?

The issue names no files, tests, or entry points. First clarify whether metrics should be pushed periodically or incremented per job change, then identify the relevant queue/job lifecycle code and define the reporting behavior and tests before implementation.

Written by the indexing model from the issue text.