Strange... I wonder if this is some integer overflow problem. There was one reported recently and fixed as of CVS head yesterday, you might try that, however without the -d2 output I'm only guessing at why pg_autovacuum is vacuuming so much / so often.
I'll see what I can do tomorrow to track it down.
I have already recommended to the program manager that they switch to 7.4.2 plus the autovacuum patch. Not sure they will be willing to make any changes at this stage in their release process though.
If we can't find one, any chance you can do some testing with CVS HEAD just to see if that works any better. I know there has been a fair amount of work done to improve this situation (not just vacuum delay, but ARC etc...)
I might do that, but not likely on Solaris. I can probably get a copy of the current database and testing scripts, and give it a try on one of my own machines (all Linux, either RHAS3, RH9, or Fedora).
Joe
---------------------------(end of broadcast)--------------------------- TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
