On Mon, 3 Nov 2008, Ivan Sergio Borgonovo wrote:
> I'm looking for a bit more guidance on gin index creation.
>
> The process:
> - vaccum analyze.
> - start a transaction that:
> - drop the triggers to update a tsvector
> - drop the index on the tsvector
> - fill several tables
> - update the tsvector in a table with ~800K records
> - recreate the gin index
> - commit
>
> To have a rough idea of the data involved:
> - 800K record
> - tsvector formed from concatenation of 6 fields
> - total length of concatenated fields ~ 200 chars *
> - average N of lexemes in tsvector 10 *
> [*] guessed
>
> 2xXeon HT 3.2GHz, 4Gb RAM, SCSI RAID5
>
> Index creation takes more than 1h.
>
> maintenance_work_mem is still untouched. What would be a good value
> to start from?
> Anything else to do to improve performances?
why you didn't change maintenance_work_mem ? You can change it online just
before create index. Bulk gin index creation uses it as a buffer and you can
save a lot of IO.
All this written in the documentation and there are other parameters you
should be concerned about.
Regards,
Oleg
_____________________________________________________________
Oleg Bartunov, Research Scientist, Head of AstroNet (www.astronet.ru),
Sternberg Astronomical Institute, Moscow University, Russia
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(495)939-16-83, +007(495)939-23-83