On Fri, Aug 09, 2013 at 02:11:46PM -0400, Tom Lane wrote:
> Cache invalidation during abort should *not* lead to any attempt to
> immediately revalidate the cache. No amount of excuses will make that
> okay. I have not looked to see just what the path of control is in this
> particular case, but we need to fix it, not paper over it.
+1. What if (sub)transaction end only manipulated the local invalidation
message queue for later processing? Actual processing would happen after
CleanupSubTransaction() returns control to the owning xact, or at the start of
the next transaction for a top-level ending.
--
Noah Misch
EnterpriseDB http://www.enterprisedb.com