metering and event time-to-live should not default to infinite

[Xav Paice]

The current charm defaults to setting the metering-time-to-live and event-time-to-live to -1, which will default to keeping all data forever. This will mean that a site is going to continually grow their mongodb usage until they run out of disk - at which point they'll need to find the config, set it, and then defrag their database after the massive performance hit from trimming old records. Perhaps it should be set to something long, but finite, to prevent this - e.g. a month?

[James Page (james-page)]

I raised this on the openstack-dev ML for further discussion, as it would be good to get some consensus as to what TTL's get used in real world deployments.