Re: vacuum full fills up disk
| От | Moritz Lennert |
|---|---|
| Тема | Re: vacuum full fills up disk |
| Дата | |
| Msg-id | fd3018b92374a95daa48ff1b91f518f5 обсуждение исходный текст |
| Ответ на | Re: vacuum full fills up disk (Tom Lane <tgl@sss.pgh.pa.us>) |
| Список | pgsql-general |
> "Moritz Lennert" <mlennert@club.worldonline.be> writes: >> I have a table with about 10 million tuples of which you can find the >> definition below. I just did an update which filled in the ss_trav_tot >> field for every tuple. Since the database will be read only (no >> updates), >> I decided to reclaim disk space by running a vacuum full. However, the >> process is slowly but surely filling up my disk. Does vacuum full make a >> copy of the data, or why is this happening ? > > It has to temporarily make duplicate index entries for each row it > moves. > > You might consider dropping the indexes, vacuum full, rebuild indexes. Thanks, I'll try that. > (You might also ask yourself if you really need all those indexes...) Well, this is data which is used for research purposes with different researchers extracting different information. I created the indexes on an intuition of which fields might be the ones information is extracted by. What other criteria might I use to determine which indexes I need ? Moritz
В списке pgsql-general по дате отправления: