Re: [AG-TECH] Problems connecting to AGServiceManager on capture node

[Michael Braitmaier <braitmaier@xxxxxxx>]

Hello Doug!

In the meantime I experienced several problems with the Fedora rpms when 
installed.
I don't know where they come from ( if they are my fault or inherit to 
the rpms), but I get somehow satisfying results
just using your AG3 dependency rpms and then completely use my AG3 CVS code.

Especially the bugs about service enabling, as well as some other 
exceptions which are thrown during running AG3
are disturbing and limiting AG's functionality.

For now I will stick with the CVS code,a s I need the nodes running by 
tomorrow evening.
Maybe we can discuss the issues in more detail on the AGRetreat, as far 
as you join?

Michael

Douglas Kosovic schrieb:

> Hi Michael,
>
> Michael Braitmaier wrote:
>
> > In the meantime the behaviour is like the following:
> >
> > I can connect to my capture node without problems:
> > Capture node:
> > video & audio capture
> > running
> > AGServiceManager -n (means NodeService also moved to this machine)
> >
> > DisplayMachine:
> > -Running VenueClient with personalNode=1
> > -Can't get displays machine ServiceManager
> > -Capture machine servicemanager available
> > -Upon "Connect to Node" I get display machines servicemanager.
>
>
> Interesting, I've never used that combination with AG3, I just start a 
> ServiceManger on the capture PC and use the NodeService that is 
> automatically started with the VenueClient on the display PC. I also 
> don't use the personalNode argument with AG3, personal node or otherwise.
>
> Since you are using straight AG CVS code on FC5, the Avahi service 
> discovery code I wrote in AccessGrid/ServiceDiscovery.py wasn't 
> designed to work using straight CVS code. In ServiceDiscovery.py 
> you'll see the following code:
>    avahiDiscoverExe = os.path.join(AGTkConfig.instance().GetBinDir(),
>                                    'ag-avahi-discover3.py')
> it won't find the ag-avahi-discover3.py file, but there is a 
> tools/ag-avahi-discover.py file in the CVS repository which needs to 
> be used instead.
>
> > In addition the VideoConsumerService seems not to be able to find vic.
> > Seems as if the vic executable is not included
> > into the VideoConsumerService-Zip. Possibly this error is restricted to
> > my local machines, as I am running from
> > CVS here, but I experience the same behaviour under FC3 with
> > VideoConsumerService.
>
>
> That's a CVS issue, the CVS code will probably need to modified to 
> fallback to /usr/bin/vic or /usr/bin/rat if it can't find one 
> extracted from the corresponding zip file (or alternatively you would 
> need to build or copy vic and rat so that the script that generates 
> the zip files finds them). For the AccessGrid 3.0 RPM, I patch the 
> code to use /usr/bin/rat and /usr/bin/vic. I'll probably checkin 
> something into CVS that makes my patch redundant.
>
> For people that are interested in running CVS code, but don't want to 
> go to the effort of doing a CVS checkout, etc. I've placed an 
> AccessGrid 3.0 CVS snapshot RPM for python 2.4 based Fedora's (i.e. 
> FC4 & FC5) here:
>   http://www.vislab.uq.edu.au/accessgrid3/fedora/development/noarch/
> and will be creating the CVS snapshots periodically.
>
>
> Cheers,
> Doug


--
------------------
Dipl.-Inf. Michael Braitmaier
HLRS - Visualization / Video Conferencing
University of Stuttgart
Germany
Phone: ++49 711 685 65996
Fax  : ++49 711 682 357
Website: http://www.hlrs.de/people/braitmaier/