Thread: Alter column varchar(n) via updating pg_attribute

Alter column varchar(n) via updating pg_attribute

From
Geoff Tolley
Date:
The last time this was asked about
(http://archives.postgresql.org/pgsql-admin/2010-02/msg00189.php), 9.0
was still alpha.

Is there any reason to suspect that direct pg_attribute manipulation to
expand a varchar(n) column might not play nice with streaming
replication/hot standby?  (Naturally I'm also in the process of testing).

Thanks,
Geoff


---------------


Re: Alter column varchar(n) via updating pg_attribute

From
Tom Lane
Date:
Geoff Tolley <geoff.tolley@yougov.com> writes:
> The last time this was asked about
> (http://archives.postgresql.org/pgsql-admin/2010-02/msg00189.php), 9.0
> was still alpha.

> Is there any reason to suspect that direct pg_attribute manipulation to
> expand a varchar(n) column might not play nice with streaming
> replication/hot standby?  (Naturally I'm also in the process of testing).

I can't see why it wouldn't.  The WAL/replication mechanism works at
such a low level that it has no idea whether an update to pg_attribute
originated in an official DDL command or a manual UPDATE of the catalog.
I think you should be more worried about fat-fingering the update than
whether it will be replicated correctly.

But, of course, testing is always a good idea ;-)

            regards, tom lane