Re: 8.3.0 upgrade, confused by documentation

Поиск
Список
Период
Сортировка
От Tena Sakai
Тема Re: 8.3.0 upgrade, confused by documentation
Дата
Msg-id FE44E0D7EAD2ED4BB2165071DB8E328C0378F07B@egcrc-ex01.egcrc.org
обсуждение исходный текст
Ответ на 8.3.0 upgrade, confused by documentation  ("Tena Sakai" <tsakai@gallo.ucsf.edu>)
Ответы Re: 8.3.0 upgrade, confused by documentation  ("Markus Bertheau" <mbertheau.pg@googlemail.com>)
Список pgsql-admin

Hello Peter,

Many thanks for your thoughts.  I appreciate it.

> It is also worth running
> ./configure --prefix /usr/local/pgsql-8.3.0
> and then make /usr/local/pgsql a symlink
> (ln -s pgsql-8.3.0 /usr/local/pgsql) then all
> you need to do to switch versions is stop
> postgres switch the symlink and restart

Excellent!

> (Very handy when doing a minor version upgrade
> and wanting to keep downtime to a minumum!
> but will not work across major versin upgrades)

Ooops!  I am going from 8.2.4 to 8.3.0 and that
*IS* a major version upgrade, isn't it?

This is all very interesting and educating to
me, but I am still stuck with the same problem.
Namely, I don't know how to use 8.3.0 to dump
data from 8.2.4 (which is shutdown in step 2, if
I am to follow the steps in the documentation).

After a long torture of my brains, I came up with
the steps below:

a) build 8.3.0 as in step 4 of the manual.
b) keep the 8.2.4 up and running, but via pg_hab.conf
   make sure nobody can use the database.
c) as user postgres, while environmental variables
   PGHOST and PGPORT is set correctly, execute the
   unix command below:
   (path_to_8.3.0)/pg_dumpall > data_saved
d) do step 2: shutdown 8.2.4
e) do step 3: mv /usr/local/pgsql /usr/local/pgsql.824
f) do step 5: create a new database cluster
g) do step 6: restore pg_hba.conf and postgresql.conf
h) do step 7: start 8.3.0
i) dp step 8: restore by:
     (path_to_8.3.0/psql -d databasename -f data_saved
j) make sure environmental variables are set correctly
   for the new database and start 8.3.0 by:
     (path_to_8.3.0)/pg_ctl restart

Where I am fuzzy is step c).  Ie., can I execute 8.3.0
program (pg_dumpall) without running 8.3.0 postgreSQL?
And will it produce what I want?

I would appreciate your thoughts/critique on what I
outlined.

Regards,

Tena Sakai
tsakai@gallo.ucsf.edu


-----Original Message-----
From: pgsql-admin-owner@postgresql.org on behalf of Peter Childs
Sent: Mon 2/11/2008 11:56 PM
Cc: pgsql-admin@postgresql.org
Subject: Re: [ADMIN] 8.3.0 upgrade, confused by documentation

On 11/02/2008, Tena Sakai <tsakai@gallo.ucsf.edu> wrote:
>
>  Hello Tom,
>
> > If your platform doesn't make it easy to
> > have two versions installed concurrently,
> > one simple way to do this is to install
> > the newer PG code on another machine and
> > do the dump across the network.
>
> I am using Dell Hardware on Redhat Enterprise
> Linux and wanting to go from 8.2.4 to 8.3.0.
> Would you please tell me where I might find
> documentation/references to run multiple versions
> of postgreSQL on one OS?
>

Put simply carry out step 4 first then return to step 1 and carry out the
rest in order skipping step 4 as you have already done that.

If you wish you could init the db somewhere else (different path after -d)
change the config to run on a different port then pipe pg_dump to psql
something like however this is a bit risky unless you understand
postgresql,conf properly....

/usr/local/pgsql/bin/psql -p 5433 -d postgres -f
`/usr/local/pgsql/bin/pg_dumpall`

or


Put simply carry out step 4 first then return to step 1 and carry out the
rest in order skipping step 4 as you have already done that. It is also
worth running ./configure --prefix /usr/local/pgsql-8.3.0 and then make
/usr/local/pgsql a symlink (ln -s pgsql-8.3.0 /usr/local/pgsql) then all you
need to do to switch versions is stop postgres switch the symlink and
restart, (Very handy when doing a minor version upgrade and wanting to keep
downtime to a minumum! but will not work across major versin upgrades)

If you wish you could init the db somewhere else (different path after -d)
change the config to run on a different port then pipe pg_dump to psql
however this is a bit risky unless you understand postgresql,conf properly
The pipe can also cause problems as the reload may take much longer than the
dump and the two together will be slower.


Peter Childs




Regards,
>
> Tena Sakai
> tsakai@gallo.ucsf.edu
>
>
> -----Original Message-----
> From: Tom Lane [mailto:tgl@sss.pgh.pa.us <tgl@sss.pgh.pa.us>]
> Sent: Mon 2/11/2008 1:39 PM
> To: Chander Ganesan
> Cc: Tena Sakai; pgsql-admin@postgresql.org
> Subject: Re: [ADMIN] 8.3.0 upgrade, confused by documentation
>
> Chander Ganesan <chander@otg-nc.com> writes:
> > Tena Sakai wrote:
> >> I am going from 8.2.4 to 8.3.0.  Does, "For best results,
> >> however, try to use the pg_dumpall command from PostgreSQL
> >> 8.3.0" apply to me?
> >>
> > Yes.  You want to use the 8.3.0 version of pg_dumpall.
>
> I don't think this is particularly critical for 8.2 to 8.3, but in cases
> where you're trying to jump multiple major versions in one step, it's
> often the case that it will go smoother with the later pg_dump.
>
> If your platform doesn't make it easy to have two versions installed
> concurrently, one simple way to do this is to install the newer PG code
> on another machine and do the dump across the network.
>
>                         regards, tom lane
>
>

В списке pgsql-admin по дате отправления:

Предыдущее
От: Peter Eisentraut
Дата:
Сообщение: Re: postgres rpms for solaris and suse
Следующее
От: "Praveen Kumar \(TUV\)"
Дата:
Сообщение: warm database.