Amanda-Users

RE: How does the amanda server identify itself to clients?

2004-02-06 15:59:45
Subject: RE: How does the amanda server identify itself to clients?
From: Joshua Baker-LePain <jlb17 AT duke DOT edu>
To: Michael Kahle <michael.kahle AT scc-wi DOT com>
Date: Fri, 6 Feb 2004 15:57:41 -0500 (EST)
On Fri, 6 Feb 2004 at 2:26pm, Michael Kahle wrote

> To spare you all the pain of reading Outlook formatted mail, this 
> message has been hand formatted using a text editor. Hope it 
> works.  :) 

Thank you.  Sincerely.

> On Friday, February 06, 2004 12:28 PM Stefan G. Weichinger wrote: 
> > Hi, Michael, 
> > on Freitag, 06. Februar 2004 at 19:13 you wrote to amanda-users: 
> > MK> I will go compile now... and weep. 
> >
> > That the debian-maintainer builds packages with localhost shows that 
> > he does not know about the problems he triggers.
> > 
> > And this is not the fault of AMANDA.
> 
> True.  I don't see how he has much of a choice.  You MUST compile for 
> the machine currently.  His way will basically work fine for a local 
> machine backup running everything in Amanda, but not much use for 
> network backups. 

Actually, this is not exactly true.  The only hostnames you specify at 
configure time are the values for --with-index-server and 
--with-tape-server.  These are simply the defaults that get applied when 
you run amrecover.  They have nothing to do with amdump, and they can be 
overridden on the amrecover command line.  The stuff that does get 
hard-coded at compile time is stuff like where the tar binary lives, which 
versions of dump are there, etc.  Within a distribution, these are 
constant, and you can safely use the packages.  Many people do, quite 
successfully.

The reason that we (the mailing list) advocate compiling your own is that 
it forces you to become familiar with where amanda will put various 
things and to understand a little bit more about how amanda works.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University