[gin-jobs] Standarized Execution Environment

[Richards, AJ (Andrew)]

Hi

As Steven points out we in the NGS are having to integrate Windows
Clusters into the service, so we do have to rethink not just policy but
testing frameworks and more. Common environment variables don't solve
all our problems at all, as each application can require enough
differences that we are looking at how we specify a common way for
launching applications that in turns takes responsibility for setting
appropriate variables

Andrew

-- 
Dr Andrew Richards
Executive Director, National Grid Service
CCLRC e-Science Centre
Rutherford Appleton Laboratory
Chilton, Didcot, Oxon
Tel: +44 (0) 1235 778029
Mobile: +44 (0)7909 875898


-----Original Message-----
From: gin-jobs-bounces at ogf.org [mailto:gin-jobs-bounces at ogf.org] On
Behalf Of Thomas Soddemann
Sent: 29 March 2007 11:34
To: gin-jobs at ogf.org
Subject: Re: [gin-jobs] Standarized Execution Environment

Hi Steven,

you are absolutely right. The intention is to gather information on how
things are handled in the different Grid project in order to have a
solid foundation for a discussion on if we want to propose a common set
of environment variables from a GIN perspective and if so, what it
should be. If you feel that this is not reflected by the text in the
Wiki let us know or just update it.

Currently, in DEISA we have only UNIX like operating system where
modules work (including Catamount). If we had to integrate e.g. Windows
Compute Cluster, we would have to rethink this policy.

Cheers,
Thomas

Steven Newhouse wrote:
> Thomas,
>
>> I have started to modify the GIN-Jobs Wiki page 
>> (https://forge.ogf.org/sf/wiki/do/viewPage/projects.gin/wiki/GINJobs)

>> to reflect your contribution.
>
> It might be worth ensuring there is a bit more community consensus 
> beyond EGEE & DEISA before making this GIN policy!
>
>> The rest is handled via the modules environment which is in the 
>> user's default path.
>
> And if you don't use modules? What if the platform is not a Unix based

> system - a scenario that has been challenging the NGS recently.
>
> Steven