Hi all,
last week ( 27/7/2003 ) I did a post with subj:
"postmaster core ( finally I have it )",
at that time I was supspecting that the core was caused by
a select on a view ( the view is always the same that cause the core )
that was running together with a vacuum; Tom Lane told me that
is really strange that vacuum could cause a core of a process doing
a select on a view; so that day I removed this normal, I think, race
( vacuum + select on that view at the same time ) and after 7 days
I still don't have a new core ( normally I already had it ).
My conclusion ( is true that I never seen the vacuum code ) is
that vacuum do something strange.
Is also true that I never seen some one have this problem but may be
because the core is not created if you don't esplicit set ulimit.
My DB is really stressed I have an average of 70 connections
and an average of 1600 commit for each minute.
Regards
Gaetano Mendola
BTW I had another strange situation in the past: a reindex on
a table was the cause of having a primary key duplicated. I removed
( now are months ) the scheduled reindex and I never seen again
the primary key duplicated.