1. (The long runnging) Long Running Job Use case
========================================================

We looked at the Long Running Job Use and the issue of proxy renewal.
Akos presented 2 sequence diagrams. The sequences 2.1 and 2.2 should
be removed (action on Akos to update the diagram), they are not
correct.The diagram in fact contains 2 use cases; Job submission and
the upload of credentials to the myProxyServer.

The location of the myProxyServer should be published in the
IS. Whoever starts up/installs the myproxy server must include a few
lines in the startup script to publish the existance of the service in
R-GMA. Steve says that the facility to do this exists already.

It was agreed that on the UI, a default myproxy server will be
configured. This is the responsibility of WP6

The proxy renewal daemon will run on the RB machine and not on the
myproxy server: 
- myproxy-init -s $MY_PROXY_SERVER

Should proxy renewal be te default ?  

 - Setting proxy renewal involves some registration overhead, it is
   best not to set it for short jobs

 - How will a user know whether proxy renewal will be required ? The
   calculation will be influenced by the ETT + time to run + time to
   replicate data and stage in data. Given that this may not be easy to
   calculate, many users will probably set the proxy renewal attribute.
  
 - Proxy renewal should not be default in all jobs since not all users
   will want to use it all in all jobs. If users want to use proxy
   renewal for their jobs, they must explicitly state this by setting
   the myproxy atribute in the JDL.

 - We want to show the proxy renewal service to work reliably first
   before setting it as the default.

Procedure to renew a proxy for a running job: 

   - job is submitted
   - when the job proxy is about to expire, the proxy renewal daemon
     performs aproxy renewal (this can be anytime during the job's
     lifetime). This is internal to the jobManager.  
   - when condorG sees that there is a new proxy, it pushes the new proxy 
     (sequence 5.2) to the job manager (via the gatekeeper)
   - this new proxy is then pushed to the WN and replaces the expired proxy
     (akos says that there is no need for an extra arrow in the
      sequence diagram cause the CE and WN are using a shared file system. )

Actions:
    * Akos and Fab:
      should update the diagram  so that this Shared file system interaction
      is obvious
    * Akos should remove the incorrect sequences from the diagram
    * Fab should present the devious flow use case 


Baseline API
============

Must be finalised. Leanne is making sure that everyone produce the
interface RPMS and tell me about them!

Web pages
==========
Leanne is the webmaster, but she WILL delegate.
  

Jeff's Metadata Issue 
======================

What is the correspondance between the SE directory structure and the
LFN ?  
    - there is now no correspondance between the LFN name and the
      path to the bytes on disk.  
    - WP9 requirements: the directory 'like' structure seems fine,
      need to check the D9.3 

Actions: 
Leanne to read D9.3 and make sure that the requirements are met by the
DM tools 
   
BrokerInfo API
======================

  The RB IS still producing the brokerinfo file.  WP2 are to support
  the brokerinfo API

    
Information Service
=======================

 It has become obvious that the IS is a lot more a part of the
 architecture than we originally realised.
 
 
Priorities until the end of the project
============================================

Prioritized lists have been produced and are attaced to the agenda
page. 

Outbound connectivity: 
Action: Erwin is to get an answer to whether we will have/use outbound
connectivty - this decision involves Cal.

Issue of LCFG 
    - it should only configure the service, but a lot of components do
       more than this 
    - check all the LCFGngs to check that they do service
      configuration only!