I think I am on to something.
The funny way I update the system tables with vacuum optimization
information is wrong, I think, and the new cache code is properly
complaining about it.
I will keep testing. If I comment out the following code, the vacuum works.
Is there a better way to do this?
---------------------------------------------------------------------------
/* XXX -- after write, should invalidate relcache in other backends */
WriteNoReleaseBuffer(ItemPointerGetBlockNumber(&rtup->t_ctid));
/*
* invalidating system relations confuses the function cache of
* pg_operator and pg_opclass, bjm
*/
if (!IsSystemRelationName(pgcform->relname.data))
RelationInvalidateHeapTuple(rd, rtup);
--
Bruce Momjian | http://www.op.net/~candle
maillist@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026