Re: Hanging with pg_restore and large objects - Mailing list pgsql-general

From Tom Lane
Subject Re: Hanging with pg_restore and large objects
Date
Msg-id 28940.1291758005@sss.pgh.pa.us
Whole thread Raw
In response to Re: Hanging with pg_restore and large objects  ("Reuven M. Lerner" <reuven@lerner.co.il>)
List pgsql-general
"Reuven M. Lerner" <reuven@lerner.co.il> writes:
> Hi, everyone.  Alban wrote:
>> Which version of pg_dump did you use? The one that came with the 9.0 install or the one from the old 8.3 one? It
shouldhave been the first of these two. 

> The dump was done by someone using the old, existing system, which runs
> under 8.3.

Hmmm ... I wonder whether this is related to the known problem that
8.3's pg_dump doesn't correctly detect file seekability under Windows:
http://archives.postgresql.org/pgsql-hackers/2010-06/msg01227.php
That means the dump file will not contain any data offset pointers.
Up to now that was only known to cause issues for parallel pg_restore,
but maybe you found another case.  But that's just a hypothesis, and a
quick test here doesn't seem to support it: I can still do pg_restore -a
from a blob-containing dump that I forced to not have data offsets.
OTOH I'm not using Windows.

Does it work any better if you use 9.0's pg_dump to dump from the 8.3
server?

            regards, tom lane

pgsql-general by date:

Previous
From: "Reuven M. Lerner"
Date:
Subject: Re: Hanging with pg_restore and large objects
Next
From: "James B. Byrne"
Date:
Subject: PG84 and SSL on CentOS-5.5 was PG84 and SELinux