Re: Page at a time index scan

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Heikki Linnakangas <hlinnaka(at)iki(dot)fi>
Cc: pgsql-patches(at)postgresql(dot)org, Simon Riggs <simon(at)2ndquadrant(dot)com>
Subject: Re: Page at a time index scan
Date: 2006-05-03 18:49:44
Message-ID: 18365.1146682184@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-patches

Heikki Linnakangas <hlinnaka(at)iki(dot)fi> writes:
> The point remains, however. A page won't get deleted while a scan
> might still be interested in it, because deleted pages are not
> immediately recycled (except on vacuum full), and the left and right
> sibling pointers stay intact until no transaction can be interested in it.

Right. AFAICS there isn't any assumption there that isn't already made
by indexscans, since we drop pin before moving to the adjacent page
anyway. You still have to be prepared to deal with the same situations.
(The new assumption is that index items won't be moved onto a
pre-existing right sibling page; the old scan logic didn't assume that.)

Heikki, were you planning to make a round of revisions in the patch,
or is this as far as you wanted to take it?

regards, tom lane

In response to

Responses

Browse pgsql-patches by date

  From Date Subject
Next Message Dhanaraj M 2006-05-04 05:19:27 current version: Patch - Have psql show current values for a sequence]
Previous Message Tom Lane 2006-05-03 18:25:54 Re: patch review, please: Autovacuum/Vacuum times via stats.