use graphite to monitor health of the system.

Collaborative tasking:
http://etherpad.wikimedia.org/p/analytics-72138

For most services, WMF mostly uses Icinga to monitor them.

To me, it seems Icinga would also be a good fit for Wikimetrics too, as
wikimetrics has many parts that are “either working or not”, and does
up to my knowledge not so much depend on performance counters going up
and down.

How comes we want to use graphite?

The reason for graphite in the title of the bug is because of the new monitoring that's been made available in labs. We assumed it was easy to add more monitoring using graphite, but we haven't looked into it.

So “graphite” is just a placeholder for $SOME_MONITORING_SERVICE?

Then just to have it written down ... in labs it seems we're getting (or
already having?) Shinken, which can consume Icingia config files IIRC.

So whoever ends up doing the Spike, might want to have look there too.

(YuviPanda would know more about Shinken.)