Re: doc fixes: vacuum_cleanup_index_scale_factor

Justin Pryzby <pryzby@telsasoft.com>

From: Justin Pryzby <pryzby@telsasoft.com>
To: Alexander Korotkov <a.korotkov@postgrespro.ru>
Cc: Robert Haas <robertmhaas@gmail.com>, Teodor Sigaev <teodor@sigaev.ru>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2018-05-08T11:05:42Z
Lists: pgsql-hackers
3rd iteration ; thanks for bearing with me.

On Tue, May 08, 2018 at 12:35:00PM +0300, Alexander Korotkov wrote:
> Hi, Justin!
> 
> Thank you for revising documentation patch.
> 
> On Mon, May 7, 2018 at 7:55 PM, Justin Pryzby <pryzby@telsasoft.com> wrote:


+        In order to detect stale index statistics, the number of total heap
+        tuples during previous statistics collection is stored in the index
+        meta-page.

Consider removing: "during previous statistics collection"
Or: "during THE previous statistics collection"

+ Once the number of inserted tuples since previous

since THE previous

+        statistics collection is more than
+        <varname>vacuum_cleanup_index_scale_factor</varname> fraction of

Since the multiplier can be greater than 1, should we say "multiple" instead of
fraction?

+        during <command>VACUUM</command> cycle.  Thus, skipping of the B-tree
+        index scan during cleanup stage is only possible when second and
+        subsequent <command>VACUUM</command> cycles detecting no dead tuples.

Change "detecting" to "detect".  Or maybe just "find"

Justin


Commits

  1. Various improvements of skipping index scan during vacuum technics

  2. Skip full index scan during cleanup of B-tree indexes when possible