Minutes of LCG GD UK/I Phone Conference , 30th September 2004

Present

Glasgow: F Speirs (minutes)
RAL: S Traylen, J Coles, C Brew, B Saunders
Cambridge: F Brochu, S Das
London T2: Owen Maroney
SouthGrid: Rhys Newman
Dublin: John Walsh
RHUL: S George
Ben Waugh


1. New changes, developments and issues from sites.

Scotgrid (F Speirs)
===================

Glasgow is at 2.2.0.  Edinburgh at 2.1.1.

Glasgow is proceeding with a plan for integrating the back-end farm based
on Scientific Linux worker nodes.

Durham will imminently be joining TestZone running 2.2.0.  They 
have a stock LCFGng installation with PBS prepared and are waiting 
for a RAID system to be repaired.

Ganglia will be deployed across ScotGrid in time.

London T2 (O Maroney)
=====================

- UCL, RH online
- Little other change from last report
- All frontends delivered, but few set up (QM, Brunel have attempted parts of installs - problems)
- Only UCL at 2.2.0, rest at 2.1.x, 2.0 backend at QMUL due to Fedora issues (recent progress on this issue)
- Upgrade to 2.2.0 planned at most sites
- SE corruption problem has been resolved by switching it to a new machine (HW fault)

Southgrid (R Newman)
====================

- All at 2.1.1 except Birmingham and RAL-PPD who are at 2.2.0
- Ganglia is well-deployed, but firewall issues at some sites
- Frontends delivered, not widely deployed
- Some work by Chris Brew on Quattor/SL ("beyond the bleeding edge")

NorthGrid (A Forti)
===================

- From email:
	- News from NorthGrid: all the sites have received the new machines. 
	- They are free to use them as more apropriate as long as the services are LCG. 
	- As for the rest there are no changes (unfortunately).
	
- All 2.1.1, ganglia

Ireland (J Walsh)
=================

- 3 Day workshop at DAIS for cluster admins
- Patched in external PBS queue to CE
- Some Globus errors in Replica Manager - trying to track down
- Some Fedora Core 2 nodes installed

Tier1/A  (S Traylen)
====================

- Tape now in operation
- Evaluating dCache



2. LCG release Status

- Imminent
- GDA Meetings involve regional update (weekly/bi-weekly) on site issues


3. GridPP head node status. 

- Some issues with network cards as reported on tb-support
- Chris Brew had success with PXE boot
- Lan1 is seen as eth0 (apparently fixed with fresher kernels)
- John Walsh reports that it might be a problem of transition between 2.4.16 and 2.4.20-30

- Rack rails are too short for 900mm racks at Oxford


4. Any Agenda Items for LCG workshop (2/3/4 November)

- Fraser mentioned discussion of site monitoring and accounting
- Jeremy requested any other updates to be emailed to him or Steve T


5. Experience of data challenges now they have stalled (as far as we are concerned).

- Feedback requested to Jeremy or Steve T.
- Owen mentioned that the release of LCG2 during DC was not good
- LHCb had done a good job helping sites get issues ironed out
- Experiment feedback from expt to sites about successful use was low

6. Support structures.

- Support (problems and followup) should come through the ROC
- This process is ongoing

7. AOB

- Question about frontends from Owen: how do sites raise support issues that are Streamline's responsibility?
- Go direct to streamline with serial numbers.

8. Next meeting 
    One month from now - Steve T will advise.