From: | Andrew Gierth <andrew(at)tao11(dot)riddles(dot)org(dot)uk> |
---|---|
To: | Mariel Cherkassky <mariel(dot)cherkassky(at)gmail(dot)com> |
Cc: | pgsql-admin(at)lists(dot)postgresql(dot)org |
Subject: | Re: size of table + toasted tables + indexes != pg_total_relation_size |
Date: | 2018-11-04 12:12:41 |
Message-ID: | 87pnvl2gki.fsf@news-spur.riddles.org.uk |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
>>>>> "Mariel" == Mariel Cherkassky <mariel(dot)cherkassky(at)gmail(dot)com> writes:
Mariel> output :
Mariel> image.png
In future please use plain text for query output, not screenshots
Mariel> When I searched for the biggest objects in DB1 I saw that the
Mariel> total sum isnt even close to 100G :
You should probably check whether there are files in the db's directory
that do not correspond to the filenode of any table (use
select pg_relation_filenode(oid) from pg_class; to get a list of
filenodes)
Crashes (or other unclean shutdown) in the middle of doing a vacuum full
or other table-rewriting operation is known to leave orphaned files
around in the data dir.
Mariel> Now, I thought that the pg_total_relation_size should include
Mariel> also all the dead tuples and the toasted tables.
It does, and your results agree with that. Perhaps you missed that
pg_table_size() includes the toast table too?
Mariel> In addition, I saw from pg_stat_activity that the session table
Mariel> has about 198 dead tuples and the toasted table has about 83833
Mariel> dead tuples. It seems that the autoanalyze has never run on the
Mariel> toasted table.
There's nothing in a toast table that needs analyzing.
--
Andrew (irc:RhodiumToad)
From | Date | Subject | |
---|---|---|---|
Next Message | Mariel Cherkassky | 2018-11-04 15:09:20 | Re: size of table + toasted tables + indexes != pg_total_relation_size |
Previous Message | Mariel Cherkassky | 2018-11-04 10:14:04 | size of table + toasted tables + indexes != pg_total_relation_size |