Re: Question about initdb -t

From: Peter Eisentraut <peter_e(at)gmx(dot)net>
To: Oliver Elphick <olly(at)lfix(dot)co(dot)uk>
Cc: PostgreSQL Development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Question about initdb -t
Date: 2000-04-02 21:18:39
Message-ID: Pine.LNX.4.21.0004021701270.373-100000@localhost.localdomain
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Oliver Elphick writes:

> The man page for initdb says that the -t option will overwrite the template1
> database with the up-to-date structure and suggests that this is the way to
> upgrade to a newer release

I believe it's wrong. The only thing -t does is fill up your template1
database with good data, say if you acidentally deleted all records from a
system table. It is not very useful generally, I think.

> How does this relate to pg_upgrade?

It doesn't. I have never used pg_upgrade but from looking at it I believe
what it does is create a new database schema (using pg_dumpall
information) and then instead of using COPY to get your data back in it
merely moves over the old on disk files for each table. The only thing
this buys you is time, it doesn't work around the various pg_dump
deficiencies or having to shut down the database, etc.

> Is it necessary to do a pg_dump?

Not unless you have something interesting in template1 you'd like to
preserve.

> If not, is there a danger of new oids in template1 overwriting old
> oids in user data?

No.

> (I am not clear whether oids are unique to a database or to the whole
> PostgreSQL installation.)

installation

--
Peter Eisentraut Sernanders väg 10:115
peter_e(at)gmx(dot)net 75262 Uppsala
http://yi.org/peter-e/ Sweden

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Peter Eisentraut 2000-04-02 21:18:55 Re: Call for porting reports
Previous Message Peter Eisentraut 2000-04-02 21:18:21 Re: Docs refreshed