... Interesting...
> Negative and positive cache entries should compete on an equal basis for
> space in the cache, since they are equally expensive to recreate.
> Can anyone spot a flaw in this reasoning?
Maybe not a flaw, but an observation: a backend running for an extended
period of time may tend to accumulate a large number of negative cache
entries and in principle they can grow indefinitely. The positive cache
entries have an upper limit of the number of actual tuples in the
catalog itself (or perhaps something smaller).
Presumably there is an upper limit to the physical cache size. Would
retaining negative entries tend to cause the cache to cycle or to grow
without bounds if there is no such limit? Or does it seem that things
would reach a reasonable steady state no matter what the query topology
tends to be?
- Thomas