Brad, okay, that makes sense.  A couple of more follow-up questions.  If
we set it up as you describe below, since they aren't replicating, that
means when a user is initially setup, his information won't flow across
to the non-replicated server.  We setup in the info field the name of
the user's Virtual Desktop Machine, and the UTWBTstart script we have
parses the other field to grab that VM name and set up the RDP session
to that host.
 
So under your scenario, we would need to periodically replicate this
info manually.  I know it can be done quite easily by dumping the
contents of utuser, utdesktop, utresadm, utresdef and utxconfig into
files, and importing them on the other server.  That or direct our
desktop support team to register a card in both failover groups when
setting up new users.
 
Also, you mentioned the inorder part of the parms file, which is fine
with 2 servers to reverse them, but we would have 3 servers in our
setup.  Two in the primary FoG and one in another FoG in a different
location.  So my understanding there is select=inorder would mean one of
the servers would never be used.  Is that correct?

________________________________

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Brad Lackey -
US-SW Desktop Product Lead
Sent: Wednesday, September 05, 2007 9:29 AM
To: SunRay-Users mailing list
Subject: RE: [SunRay-Users] Multicast problems


The issue is that if you set up a FOG across the WAN, you will be
subject to the load balancing mechanism deciding where your sessions
will live.

For example... We have two sites Site A and Site B These sites have
100MB local LAN's but are connected by a 4MB WAN link. There is one Sun
Ray server at each location, and 10 DTUs at each location.

If you set up the servers at both sites in a FOG, DTUs from each site
may have sessions on the other site during normal operation. This
wouldn't be the best use of your WAN link bandwidth.

What most customers want is their DTUS to connect to a local Sun Ray
server, and if in the case that server is down, connect across the WAN
to the alternate.

There is no Need to put these in a FOG to do this. The two things that a
FOG does is Replication of policy/registrations and load balancing.

Your best bet is to use your configuration mechanisms to give the DTUs
BOTH servers as connection targets in the preferred order.

For example, put both your servers in the servers= line of your .parms
file and set select=inorder. The servers= lines will be opposite on both
of your servers.

Brad

On Wed, 2007-09-05 at 09:28 -0400, Lewick, Taylor wrote: 

        If I understand it correctly, SUN uses AMGH, and there is a good
blog of
        an "AMGH How To Guide" here...
        http://blogs.sun.com/bobd/
        
        My problem is I'd like to setup failover groups across the WAN.
But
        according to our network guys, we don't route mutlicast across
the WAN.
        And AMGH still says its not currently supported for CAM/Kiosk
mode.  It
        can be done, but requires a little customization.  Myself, I
have no
        problem trying to do the customization, but as the only onsite
Unix
        Admin, and due to corporate requirements, I'd really rather not
start
        down the path of running in an "unsupported" configuration. 
        
        So I'm a bit stuck as to how I should go about implementing
failover
        groups across the WAN.
        
        Any ideas?
        
        Thanks,
        Taylor
        
        _______________________________________________
        SunRay-Users mailing list
        [email protected]
        http://www.filibeto.org/mailman/listinfo/sunray-users

        Brad Lackey
Desktop Product Lead
US Software Practice
(720) 548-3339
[EMAIL PROTECTED]


        

<<sun_sig2.gif>>

_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users

Reply via email to