Re: restore a pg_dumpall only breaks

Поиск
Список
Период
Сортировка
От Marc Fromm
Тема Re: restore a pg_dumpall only breaks
Дата
Msg-id CY1PR0201MB143628C631F61E62D49B86CC9DB60@CY1PR0201MB1436.namprd02.prod.outlook.com
обсуждение исходный текст
Ответ на Re: restore a pg_dumpall only breaks  (Tom Lane <tgl@sss.pgh.pa.us>)
Список pgsql-admin
> If you're working with locally-built copies of Postgres, I'd say just follow the recipe given in the pg_upgrade man
page. If you are working with someone's packaged version of Postgres, the packager may have provided a script or
somethingfor upgrades, in which case follow their recommendation.  Either way it's a good idea to make a backup
beforehandin case of disaster. 

Postgresql was installed during the install of centOS 6.4. Postgresql was an install option. Thus I figure I am working
with"someone's packaged version.  

No need for database backup since the backups are from a different server (9.2), which has prompted all of this. If I
copythe entire pgsql directory and thing blow up can I just replace the copied directory with its contents and be
restoredto 8.4?  

I downloaded pgdg-centos92-9.2-8.noarch.rpm for CentOS6.4 from https://yum.postgresql.org/repopackages.php#pg92. Is
theremore involved then using yum to uninstall 8.4 and then use yum to install 9.2? 

Thanks

-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Thursday, November 24, 2016 10:22 AM
To: Marc Fromm <Marc.Fromm@wwu.edu>
Cc: pgsql-admin@postgresql.org
Subject: Re: [ADMIN] restore a pg_dumpall only breaks

Marc Fromm <Marc.Fromm@wwu.edu> writes:
>> A dump from 9.2 is no sure thing to restore into an 8.1 database; there may be SQL syntax in it that 8.1 doesn't
understand. Have you checked for errors during the restore? 

> After the last successful database restore from the pg_dumpall file this is displayed on the ssh session.
> psql:pg_dbs.bkp:1029100: \connect: invalid connection option "-reuse-previous"

Ah.  That option was introduced quite recently as part of a security fix.
It's no surprise 8.x doesn't recognize it.

You could probably work around this by using pg_dumpall with -g to just dump roles and tablespaces, and then
pg_dump'ingindividual databases separately.  A mite tedious and error-prone.  Or maybe use "sed" to strip out the
-reuse-previousoptions from pg_dumpall's output.  But really this is the best answer, because 8.x is long out of
support:

> I figure the best thing to do is to upgrade postgres 8.4 (not 8.1 as previously mentioned) to 9.2 on this centOS6.4
server,so it's version matches the other server. I searched for some time on how to do this and every tutorial does it
alittle different and each has different steps that the others do not have. Is there a clear step by step procedure to
dothis upgrade? 

If you're working with locally-built copies of Postgres, I'd say just follow the recipe given in the pg_upgrade man
page. If you are working with someone's packaged version of Postgres, the packager may have provided a script or
somethingfor upgrades, in which case follow their recommendation.  Either way it's a good idea to make a backup
beforehandin case of disaster. 

            regards, tom lane


В списке pgsql-admin по дате отправления:

Предыдущее
От: Alvaro Herrera
Дата:
Сообщение: Re: Replica lag, high read IO, vacuum index scanning bug?
Следующее
От: Marc Fromm
Дата:
Сообщение: Re: restore a pg_dumpall only breaks