Re: trouble restoring database

On Tue, Aug 24, 2010 at 3:22 PM, Tena Sakai <tsakai(at)gallo(dot)ucsf(dot)edu> wrote:
> Having said that, I think Ian is correct that I was missing -c flag in the
> command below:
>  gunzip ..../dumpall20100822.0.gz | /usr/local/pgsql/bin/psql -f - postgres

Quick tip, you can use zcat to get the output of a gzipped file like:

zcat ..../dumpall20100822.0.gz|psql

There's also zgrep, zless, zegrep, zxpdf, and zdiff. All do what it
looks like they do.

> Since the file got uncompressed, I am now running the command:
>  psql -f /home/tsakai/Notes/postgres/dumpall20100822.0
> and I think it is working.  It is giving messsages:
>  ERROR:  role "postgres" already exists
>  STATEMENT:  CREATE ROLE postgres;
>  ERROR:  database "testdb" already exists
>  STATEMENT:  CREATE DATABASE testdb WITH TEMPLATE = template0 OWNER =
> postgres ENCODING = 'UTF8';
>  LOG:  checkpoints are occurring too frequently (5 seconds apart)
>  HINT:  Consider increasing the configuration parameter
> "checkpoint_segments".
>
>
> The last 2 lines are repeated many, many times.  This is not terribly
> Serious, is it?  (I will fix it via postgresql.conf file shortly.)

No. It's just telling you that from a performance perspective things
would be faster with more checkpoint segments. Unless you spend a
fair portion of your day restoring dbs, it's probably not big deal.
If your machines only do a little writing at a time then you can
probably leave it. However, increasing checkpoint segments does help
a lot if you do handle a fair bit of writes, and it's cheap and easy
to do.