Re: Page at a time index scan
От | Simon Riggs |
---|---|
Тема | Re: Page at a time index scan |
Дата | |
Msg-id | 1146678722.449.196.camel@localhost.localdomain обсуждение исходный текст |
Ответ на | Re: Page at a time index scan (Tom Lane <tgl@sss.pgh.pa.us>) |
Список | pgsql-patches |
On Wed, 2006-05-03 at 13:39 -0400, Tom Lane wrote: > Simon Riggs <simon@2ndquadrant.com> writes: > > So do you see a problem scenario like this? > > > A, B and C separate backends: > > A1 Reads page, some row versions are *not* marked LP_DELETE but will be > > later when A2 happens > > B1 VACUUM removes dead rows, just happens to be all of them > > B2 Recycles page into FSM > > C1 Inserts new data into old page > > A2 Attempts to update old page to notify about dead rows (UGH!) > > Can't happen; a page cannot be recycled until all concurrent > transactions are gone. In any case, the LP_DELETE marking code will > certainly take care to check that the entries it's trying to mark > are still the same ones it meant to mark. ! So do you see a problem with page deletion, or not? If so, what is it? This patch looks good to me, based upon everything said. -- Simon Riggs EnterpriseDB http://www.enterprisedb.com
В списке pgsql-patches по дате отправления: