Hi All,
I would have to agree with Karl's summation.
I would also fully endorse the symlink method, I've used it extensively on UNIX systems and never had a problem. I think there is a reference somewhere in the manual that says its not supported, but I've been told by at least one tech "Yeah and if we stopped supporting symlinks we'd be lynched!". I would also add that as you'll need root privilege to set it up in advance that your systems guys will at least know of it.
As for editing the config file...Down that path lies madness. It can be done, and I can use Perl programs to create a class of editing routines to do the job, but as I've just found out during a recovery that the journal files also contain hardcoded paths used for file creates. I'm still investigating the source of these, but it means that to recover a database using the Config file edit method will also involve a dump/journal file editor as well.
I should point out that the only reason I'm looking at this is that I have been forced into reconsidering this recovery technique by a particularly large database on an old installation that can't be upgraded.
Marty
-----Original Message-----
From: Karl Schendel [mailto:schendel (AT) kbcomputer (DOT) com]
Sent: 28 November 2011 20:48
To: Ingres and related product discussion forum
Subject: Re: [Info-Ingres] Recover checkpoint to server with different locations
On Nov 28, 2011, at 1:07 PM, Ingres Forums wrote:
Quote:
Hi,
We're performing a disaster recovery test and need to recover a
checkpoint from our production server over to our DR box. The issue I
believe I'm having is that the two servers do not have the same INGRES
data locations configured. I think we have the same number of them,
but they're not the same folder paths.
How can I make this happen? |
Andre's suggestion of making a second installation identical to production
is a good one. Or, along the same lines, you could unload the existing
database on the DR box, reinstall Ingres so that it looks identical to the
production installation, and then reload the DR database.
The other options that occur to me at the moment are less good:
You could create symbolic links to match the production installation
paths, pointing to the corresponding DR-installation paths. I can't
offhand think why this wouldn't work, but it would certainly confuse
your sysadmins and future DBA's.
You could grab a copy of the Ingres source and use the structure
definitions found in $ING_SRC/back/dmf/hdr/dm0c.h to write
a little C program that alters the paths in the db-config file(s).
It wouldn't have to be C, I guess. You would have to find and
alter all of the copies of the config file. Marty is apparently
doing something like this, since he's posted recently on the
subject. I actually did something like this in a past life for an
automated DR / warm-backup system, and it can be done,
but it's tricky to get right. I suspect it would be a lot less work to
simply redo your DR ingres installation to match production.
Karl
_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://ext-cando.kettleriverconsulti...fo/info-ingres