Re: [Proposal] Fully WAL logged CREATE DATABASE - No Checkpoints
От | Robert Haas |
---|---|
Тема | Re: [Proposal] Fully WAL logged CREATE DATABASE - No Checkpoints |
Дата | |
Msg-id | CA+TgmoY89CD8dbDF-+XY=aE6E6Vgd+Bz1scj-m5qL31nsVeNNQ@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: [Proposal] Fully WAL logged CREATE DATABASE - No Checkpoints (Bruce Momjian <bruce@momjian.us>) |
Ответы |
Re: [Proposal] Fully WAL logged CREATE DATABASE - No Checkpoints
|
Список | pgsql-hackers |
On Mon, Feb 14, 2022 at 1:58 PM Bruce Momjian <bruce@momjian.us> wrote: > > I think we have consensus on STRATEGY. I'm not sure if we have > > consensus on what the option values should be. If we had an option to > > use fs-based cloning, that would also need to issue a checkpoint, > > which makes me think that CHECKPOINT is not the best name. > > I think if we want LOG, it has tob e WAL_LOG instead of just LOG. Was > there discussion that the user _has_ to specify and option instead of > using a default? That doesn't seem good. I agree. I think we can set a default, which can be either whatever we think will be best on average, or maybe it can be conditional based on the database size or something. -- Robert Haas EDB: http://www.enterprisedb.com
В списке pgsql-hackers по дате отправления: