Re: Lost replication slots after pg_upgrade.

Поиск
Список
Период
Сортировка
От Victor Sudakov
Тема Re: Lost replication slots after pg_upgrade.
Дата
Msg-id YfuDt1LU4N0Wx8rR@admin.sibptus.ru
обсуждение исходный текст
Ответ на Re: Lost replication slots after pg_upgrade.  (Vijaykumar Jain <vijaykumarjain.github@gmail.com>)
Список pgsql-admin
Vijaykumar Jain wrote:
> On Thu, Feb 3, 2022, 10:54 AM Victor Sudakov <vas@sibptus.ru> wrote:
> 
> > Nikhil Shetty wrote:
> > > Hi,
> > > I have a Primary and Standby running on Postgres v11.7.
> > >
> > > I was following the document to upgrade a PG v11.7 cluster to PG v13.4.
> > > After upgrade using pg_upgrade on primary, I ran rsync to setup standby
> > > PG13 but when I started the primary, the replication slots were not
> > > present. How will standby start streaming in this case?
> > >
> > > We are using physical replication slots.
> >
> > You can create the replication slots manually with
> > pg_create_physical_replication_slot() any time.
> >
> > However, unless you are very brave and know what you are doing,  I'd
> > recommend using pg_basebackup to setup a standby, not rsync. You can
> > even use the -C and -S options of pg_basebackup to create the slots
> > for you, and many other pg_basebackup's nice
> >
> 
> A small demo of 11 to 13 upgrade using pg_upgrade and rsync.
> It might not be all the best practices, but just following the doc on my
> laptop.
> 
> >
> https://gist.github.com/cabecada/4517af13245383b888833cfc69d741be
> 
> Just ensure the backed up configs are in back in and also create the
> replication slot.

The black magick of using rsync instead of pg_basebackup is even
documented in https://www.postgresql.org/docs/13/pgupgrade.html but
it's much more difficult and error-prone than using pg_basebackup. 

-- 
Victor Sudakov VAS4-RIPE
http://vas.tomsk.ru/
2:5005/49@fidonet



В списке pgsql-admin по дате отправления:

Предыдущее
От: Vijaykumar Jain
Дата:
Сообщение: Re: Lost replication slots after pg_upgrade.
Следующее
От: Peter Eisentraut
Дата:
Сообщение: Re: pgbouncer logs and its efficiency