shibboleth-dev - RE: shib machine requirements...... final call
Subject: Shibboleth Developers
List archive
- From:
- To: <>
- Subject: RE: shib machine requirements...... final call
- Date: Fri, 21 Dec 2001 09:11:22 -0500
(for "Mike LaHaye"
<>):
Steven,
My read on this is that we need:
2 Linux Servers for Development/Testing
(minimally a dual 550 MHz PIII with 18 GB disk, 512MB RAM, FE)
- Redhat 7.2
- Openldap 2.0.11 or better
- Java SDK 1.3.1
- Apache 1.3.22
- mod_ssl and OpenSSL (whatever's recent; 0.9.6b being used at cmu)
- Tomcat 3.3
- the mod_jk plugin for apache
- gcc/g++ 2.95.3
- pub-cookie (http://www.washington.edu/computing/pubcookie/ext/)
- Mysql 3.23.40 - 3.23.44
- Perl 5.6.0
* Please comment on an additional hardware requirements.
2 Solaris Servers
- Solaris 2.8
- iPlanet directory server
- Java SDK 1.3.1
- Apache 1.3.22
- mod_ssl and OpenSSL (whatever's recent; 0.9.6b being used at cmu)
- Tomcat 3.3
- the mod_jk plugin for apache
- gcc/g++ 2.95.3
- pub-cookie
- Mysql 3.23.40 - 3.23.44
- Perl 5.6.0
CVS service available
We can build dedicated Linux boxes for this effort. On the Solaris side,
we have a pair of Sun 220R servers for our internal deployment of
iPlanet. I'm hoping that we can leverage these servers for demonstration
purposes of that platform. This might be optimistic. Comments?
If we need dedicated Solaris boxes for this effort, will the Netra X1
server meet the need?
Mike LaHaye
Internet2
-----Original Message-------
From:
[mailto:]
Sent: Wednesday, December 19, 2001 10:57 AM
To:
Cc: Mike LaHaye
Subject: Re: shib machine requirements...... final call
At 3:54 PM -0500 12/18/01, Russell J. Yount wrote:
>Steven,
> Different architectures for production verses development is
>less that optimal.
>If we need to track down some alignment or byte order problems it
>would best to have
>one of each type for both development and production. Or just stay
>with one acrhitecture
>for now.
well, my take on this is that we're going to encounter these
different versions in the field, as we deploy. Do we want that
initial encounter to be in-house, in our test environments, or on
some campus (in the UK?) I don't think we're going to be able to
require that a pilot site run OpenLdap,,, especially if they already
have iPlanet installed and running. My experience tells me that
-- you're right, we're going to encounter obscure problems as a
result of having to support a variety of backends.
-- its going to be a *lot* easier to debug these problems in-house,
in a controlled environment, than at a pilot site, with different
time pressures, and an environment we don't control.
Does this make any sense?
>
> We would prefer Redhat Linux.
>
> Stock CVS would be fine.
>
> We would prefer Openldap, we do not have Netscape directory.
>
>-Russ
--
------------------------------------------------------mace-shib-design-+
For list utilities, archives, subscribe, unsubscribe, etc. please visit the
ListProc web interface at
http://archives.internet2.edu/
------------------------------------------------------mace-shib-design--
- shib machine requirements...... final call, Steven_Carmody, 12/18/2001
- RE: shib machine requirements...... final call, Scott Cantor, 12/18/2001
- <Possible follow-up(s)>
- Re: shib machine requirements...... final call, Steven_Carmody, 12/19/2001
- RE: shib machine requirements...... final call, Steven_Carmody, 12/21/2001
- RE: shib machine requirements...... final call, Steven_Carmody, 12/21/2001
Archive powered by MHonArc 2.6.16.