How to add fluentd liveness probe in OCP 3

Issue

• Currently it looks like the Fluentd daemonset that comes out of the box isn't configured with a liveness probe and we have seen instances where the pods have become stuck and stop working. This is usually resolved with a cycling of the pods but it would be better to have a liveness probe to automatically do this rather than a manual check.
• Is there a set of recommended/supported liveness probes to implement across the clusters?