Re: Brain dump: btree collapsing
От | Daniel Kalchev |
---|---|
Тема | Re: Brain dump: btree collapsing |
Дата | |
Msg-id | 200302130816.h1D8GaY18052@dcave.digsys.bg обсуждение исходный текст |
Ответ на | Re: Brain dump: btree collapsing (Justin Clift <justin@postgresql.org>) |
Список | pgsql-hackers |
>>>Justin Clift said:> <snip>> > In theory, if we find recyclable page(s) at the physical end of the index,> > we could truncatethe file (ie, give the space back to the filesystem)> > instead of reporting these pages to FSM. I am not sure ifthis is worth> > doing --- in most cases it's likely that little space can be released this> > way, and there may be sometricky locking issues.> > Sounds like this would be beneficial for environments with high > update/delete transactionvolumes, perhaps on smaller amounts of > live/valid data. But if dead pages are removed (returned to FSM?) as soon as last item is removed from the page, the page usage will remain small. Or perhaps not? That is, it's unlikely to collect large number of dead/free pages at the end of the physical storage except if doing all this in single VACUUM session. Daniel
В списке pgsql-hackers по дате отправления: