Re: Skipping logical replication transactions on subscriber side
| От | Amit Kapila |
|---|---|
| Тема | Re: Skipping logical replication transactions on subscriber side |
| Дата | |
| Msg-id | CAA4eK1+woM2cWGkDzy7JFSze4d8gJy7uBZdyAJVS7pSgcyb+-g@mail.gmail.com обсуждение исходный текст |
| Ответ на | Re: Skipping logical replication transactions on subscriber side (Masahiko Sawada <sawada.mshk@gmail.com>) |
| Список | pgsql-hackers |
On Thu, Jul 1, 2021 at 1:24 PM Masahiko Sawada <sawada.mshk@gmail.com> wrote: > > On Wed, Jun 30, 2021 at 8:05 PM Amit Kapila <amit.kapila16@gmail.com> wrote: > > > If so, in which cases are you imagining the files to be > > created, is it in the case of relation message > > (LOGICAL_REP_MSG_RELATION)? Assuming the previous two statements are > > correct, this will skip the relation message as well as part of the > > removal of stream files which might lead to a problem because the > > publisher won't know that we have skipped the relation message and it > > won't send it again. This can cause problems while processing the next > > messages. > > Good point. In the current patch, we skip all streamed changes at > stream_commit/abort but it should apply changes while skipping only > data-modification changes as we do for non-stream changes. > Right. -- With Regards, Amit Kapila.
В списке pgsql-hackers по дате отправления: