I (finally) realized that my script for ANALYZEing parents of table hierarchies
every month or so was looping around the same parent tables every night due to
no stats for date of last last analysis.
I guess that's deliberate/known and maybe related to relkind='p' having no
relfilenode.
Is there any good workaround other than making stampfiles or making my own
"last analyzed" table?
Thanks,
Justin