On Sat, Jun 6, 2015 at 11:51 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> This suggests that CLOBBER_CACHE_ALWAYS is actually missing a pretty
> large part of the cache behavioral space. Maybe we should devise some
> sort of "CLOBBER_CACHE_RANDOMLY" option that would inject cache flush
> events more selectively, perhaps only once every thousand opportunities
> or so. And perhaps not only full cache reset events, though I confess
> to not being sure what that ought to look like.
It should probably be done pseudo-randomly, but otherwise, +1.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company