I am doing some testing on memory consumption for term facets. I ran a test
where I index some content and calculate a terms facet on it. I am using
the node stats request to learn about the field cache size (after
calculating the facet). My expectation was that I see different memory
consumption when I go from string to integer or even byte fields. The
strange thing is that the field cache size reported by the node stats is
always the same (regardless of the field value type being faceted). Does
the stats (maybe) only measure the field cache array size (without the
actual field values)?
--