Hi,
apologies to the people who were on the phone. We were online too but
there was only music.
Present: A. Forti M. Houlden M. George P. Hodgson
> a) technical issues at each sites.
Most of the site didn't have any major technical problem to report.
However since sites now belong to a common entity which has to answer for
each sites that doesn't work properly two situations have been identified
that need to be notified to the technical coordinator or to the
northgrid-sysadmin
1) Any technical problems that cannot be solved locally in the short term
(2 days)
2) When the local sys admin is away and the site is not covered
> b) what to do with the gridpp machines.
Sites are free to introduce the new machines in the most natural way
for them. They are not forced to replace their working CE SE UI and LCFG
In particular since LCFG is not going to be used anymore one of the
machines can become a quattor test machine.
> c) alternative to lcfg for the installation.
Liverpool is developing software to replicate nodes images. This allows to
install 300? machines in 20 minutes. This is a really interesting way of
installing and upgrading but since the software is under development the
decision is to start to play with quattor (Manchester is already committed
to this) and when the Liverpool software is available to evaluate it. The
two requirement the Liverpool software has to satisfy before being
avaluated:
1) it has to be user friendly
2) Liverpool has to garantuee to maintain it
> d) northgrid services do we want to install centralised or distributed
> services for northgrid? RB, BDII.
The idea of having a NorthGrid RB was considered technically a good idea
but since it might have potential political issues it has also to be
discussed and approved by the management board. We probably have to
monitor CERN and RAL down times.
> h) ganglia installation
It was agreed to have a ganglia server in Manchester to group the
Tier2 sites, it was also agreed that it is not a good idea to have a
uniform page design that identifies a NorthGrid site. Sites behind a
firewall have to open it.
> e) organisation of future meetings.
For the time being we will have a meeting once a month, we will pass to
once every three months possibly before gridpp meetings when everything
will be considered working.
> f) any issue you might want to raise.
Shared resources:
- there is no way to flag a cluster or part of it
as "suspended because used by co-owners" on the GOC database and
at testZone level. Alessandra will start to raise the issue.
- The new type of monitoring is going to read some of the machines
log files. The method is highly invasive and should be questioned
both for local and shared resources usage.
Both the points have political relevance and should be discussed
by the Management Board.
Allowing non officially supported VOs to use resources
- It was considered in the spirit of the grid and it raises the
efficency of the equipement from an average 50-60% or lower to
80-90%
This should be kept in consideration by the Management Board when
discussing this issue with the universities.
Please feel free to add and comment.
cheers
alessandra
|