Re: [GENERAL] bloat indexes - opinion - Mailing list pgsql-general

From Patrick B
Subject Re: [GENERAL] bloat indexes - opinion
Date
Msg-id CAJNY3itbx2SMt3gpTsF2vhdjocQPJajNAje7Q5BGA9Nty08eWw@mail.gmail.com
Whole thread Raw
In response to Re: [GENERAL] bloat indexes - opinion  (Adrian Klaver <adrian.klaver@aklaver.com>)
Responses Re: [GENERAL] bloat indexes - opinion  (Patrick B <patrickbakerbr@gmail.com>)
List pgsql-general
2017-02-22 10:59 GMT+13:00 Adrian Klaver <adrian.klaver@aklaver.com>:
On 02/21/2017 01:44 PM, Patrick B wrote:
> Hi guys,
>
> I've got a lot of bloat indexes on my 4TB database.
>
> Let's take this example:
>
>     Table: seg
>     Index: ix_filter_by_tree
>     Times_used: 1018082183
>     Table_size: 18 GB -- wrong. The table is mostly on pg_toast table.
>     Its real size is 2TB

How do you know one number is right and the other is wrong?


1. on that table (seg) i store binary data. It is impossible to have only 18GB of it.
2. 
SELECT schema_name, 
       pg_size_pretty(sum(table_size)::bigint),
       (sum(table_size) / pg_database_size(current_database())) * 100
FROM (
  SELECT pg_catalog.pg_namespace.nspname as schema_name,
         pg_relation_size(pg_catalog.pg_class.oid) as table_size
  FROM   pg_catalog.pg_class
     JOIN pg_catalog.pg_namespace ON relnamespace = pg_catalog.pg_namespace.oid
) t
GROUP BY schema_name
ORDER BY schema_name

pg_toast 2706 GB 82.62112838877240860000 <-- this belongs to the seg table.

 

Have you looked at the functions here?:
https://www.postgresql.org/docs/9.6/static/functions-admin.html#FUNCTIONS-ADMIN-DBOBJECT

>     Index_size: 17 GB
>     Num_writes 16245023
>     Index definition: CREATE INDEX ix_filter_by_tree ON seg USING btree
>     (full_path varchar_pattern_ops) WHERE (full_path IS NOT NULL)
>
>
>
> What is the real impact of a bloat index? If I reindex it, queries will
> be faster?
>
> Thanks
> Patrick


--
Adrian Klaver
adrian.klaver@aklaver.com

pgsql-general by date:

Previous
From: Adrian Klaver
Date:
Subject: Re: [GENERAL] bloat indexes - opinion
Next
From: Rob Brucks
Date:
Subject: Re: [GENERAL] NOTIFY command impact