Re: vacuum, performance, and MVCC

Jan Wieck wrote:
> On 6/25/2006 10:12 PM, Bruce Momjian wrote:
> > When you are using the update chaining, you can't mark that index row as
> > dead because it actually points to more than one row on the page, some
> > are non-visible, some are visible.
> 
> Back up the truck ... you mean in the current code base we have heap 
> tuples that are visible in index scans because of heap tuple chaining 
> but without index tuples pointing directly at them?

No, this would be new code added.  The basic idea is with the new
same-page update chaining, a single index points to the head of a chain,
not to a single tuple, so you can't mark a tuple as pointing to dead
rows if any of the tuples in the chain are visible.

--  Bruce Momjian   bruce@momjian.us EnterpriseDB    http://www.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +