> "Maksym Boguk" <maxim.boguk@gmail.com> writes:
>> Somehow pg_database_size producing results which are 2=D1=85 more then r=
eality.
>
> I can't reproduce any such problem here. =C2=A0I suspect you overlooked s=
ome
> tablespaces in your manual "du" commands.
>
> =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=
=A0 =C2=A0regards, tom lane
>
No other tablespaces exists as can be seen in output of:
SELECT oid,spcname,spclocation,pg_size_pretty(pg_tablespace_size(spcname)) =
from
pg_tablespace;
oid | spcname | spclocation | pg_size_pretty
-------+------------+---------------+----------------
1663 | pg_default | | 21 MB
1664 | pg_global | | 1092 kB
16443 | main | /db/base/main | 124 GB
However I seem to have found reason for such strange behavior.
That is a duplicating link in pg_tblspc:
postgres@db:~$ ls -la /var/lib/postgresql/9.0/main/pg_tblspc/
total 8
drwx------ 2 postgres postgres 4096 Jun 3 04:08 .
drwx------ 12 postgres postgres 4096 Sep 28 17:08 ..
lrwxrwxrwx 1 postgres postgres 13 Jun 6 14:39 16384 -> /db/base/main
lrwxrwxrwx 1 postgres postgres 13 Jun 3 04:08 16443 -> /db/base/main
I have no idea how could this happen.
No crashes/power-offs and so on happens on that database for more then year.
Can the second (16384 -> /db/base/main) link be safely deleted?
--=20
Maxim Boguk
Senior Postgresql DBA.