GENIUS Meeting 1 Minutes

From RealityGrid

GENIUS Kick Off Meeting 5/7/07 16:00 BST

Attendees:

Peter Coveney, Stefan Zasada, Steven Manos, Marco Mazzeo, Jon MacLaren, Ashley Pittman, David Salmon, Owain Kenway, Dave Plummer, Stefan Brew, Simon Clifford, Stephen Pickles, Matt Ford, Rob Haines, Robin Pinning, Dan Katz, Shantenu Jha, Sergiu Sanielevici, Shawn Brown, Brian Toonen, Bruce Boghosian, Steven Young, Matteo Turilli, Andrew Richards, Stephen Booth, Peter Clarke.

Points arising from the meeting:

1) There are important networking requirements (JANET lightpaths) to be requested between Oxford and Manchester, and from both to UCL.

2) Steven Manos will act as liaison officer between GENIUS, UKERNA (David Salmon) and those responsible for networking provision at the the UK NGS2 nodes (whose details he now needs to identify very quickly).

3) Sean Brown @ PSC will act as a TeraGrid contanct point.

4) the LONI grid is in a state of transition, with many new resources coming online, including Queen Bee and many smaller linux clusters.

5) The GENIUS project will target the Manchester and Oxford NGS2 nodes initially, requesting that NGS installed the required software on those machines, and working with DS on the provisioning of lightpath connection.

6) PC suggests that, based on previous experience, weekly meetings should be held between people working on networking provision, to avoid serial delays getting connections in. People from relevant networking depts in partner institutions also need to be involved.

7) RP to be given access to HemeLB CVS, and to work with MM deploying the code on the NGS, and at SDSC & NCSA.

8) SB noted that the timescale for availablitiy of Cray linux is to be accelerated, meaning that accces to back end HECToR nodes may be possible. EPCC will work on getting Globus etc deployed on HECToR.

9) HARC is currently deployed on 3 LONI machines. There are 5 HARC acceptors deployed to talk to TG and LONI machines, which can also be used with RMs in the UK.

10) SS suggests when requesting TG time on upcoming DAC/MRAC applications, mention the need for HARC and SPRUCE. Applications can now also be used to request advanced support time from TG to port applications etc.

11) BT to put pointer to MPI-g info on GENIUS wiki. Currently a bug is slowing the release of MPI-g. BT advises to hold off deploying until myrinet bug is fixed. SP will give BT access to the NGS to test MPI-g.

12) RH will be point of contact for ReG steering.