Re: [ADMIN] pg_class reltuples/relpages not updated by autovacuum/vacuum
Cédric Villemain <cedric.villemain.debian@gmail.com>
From: Cédric Villemain <cedric.villemain.debian@gmail.com>
To: Alvaro Herrera <alvherre@commandprompt.com>
Cc: Kevin Grittner <kevin.grittner@wicourts.gov>,
Tom Lane <tgl@sss.pgh.pa.us>, Florian Helmberger <fh@25th-floor.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2011-05-25T19:14:06Z
Lists: pgsql-hackers
2011/5/25 Alvaro Herrera <alvherre@commandprompt.com>: > Excerpts from Cédric Villemain's message of mié may 25 13:24:01 -0400 2011: > >> > Well, we only actually need to store one number, because you can figure >> > out a much more precise number-of-pages figure with pg_relation_size() >> > divided by configured page size. > >> I may miss something but we need relation size in costsize.c even if >> we have a reldensity (or we need a reltuples). Else what values should >> be used to estimate the relation size ? (pg_relation_size() goes down >> to kernel/fs to ask the stat.st_size, is it really what we want?) > > Actually yes, the planner does go to kernel to determine the current > relation size, and then multiplies by density as computed from catalog > data to figure out current reasonably accurate number of tuples. Okay! I just read that part. Interesting. (If I dive correctly, we search our last segment and then use a fileseek to the end of this segment to get our information) make more sense, suddendly :) > > -- > Álvaro Herrera <alvherre@commandprompt.com> > The PostgreSQL Company - Command Prompt, Inc. > PostgreSQL Replication, Consulting, Custom Development, 24x7 support > -- Cédric Villemain 2ndQuadrant http://2ndQuadrant.fr/ PostgreSQL : Expertise, Formation et Support