CERN/FOCUS 2001-002
21-May-2001
Present:
J.Altaber*), T.Cass, M.Cattaneo (Secretary), M.Delfino, M.Ernst, F.Etienne,
B.Gobbo, R.Gokieli, A.Grant, F.Hemmer, H.-F.Hoffmann, V.Innocente, S.Jarp*),
P.Jeffreys (Chairperson), M.Kienzle, J.Knobloch, W.Lerche, L.Mapelli*),
M.Marquina, N.McCubbin, H.Meinhard, S.O’Neale, M.Pimiä*), H.Renshall,
L.Robertson, K.Safarik, A.Sandoval, J.Shiers, A.Silverman*), P.Vande Vyvre,
R.Voss
Invited:
S.Bethke, F.Carminati, D.Heagerty, M.Pepe, E.McIntosh, I.McLaren,
M.Stavrianakou,
Apologies:
J.Boucrot, R.Cashmore, A.Norton, W.Von Rueden
Absent:
P.Anderssen, F.Gagliardi, D.Jacobs, J.May, M.Mazzucato, E.Valente
*)
part time
1.1 Consideration
of Agenda
1.2 Minutes from last meeting and matters arising
1.3 Chairman’s
comments
2.1
Conclusions of LHC computing review
and implications for FOCUS (S.
Bethke)
2.2
Creation of Data Management group
(F.Carminati)
3. Experiment Requirements for CVS and build servers (M. Stavrianakou)
4. Issues concerning LEP analysis until 2003
4.1 Introduction (H.Renshall)
4.2 Requirements for FATMEN (S.O’Neale)
4.3 Status of migration to Linux
·
Aleph
(H.Renshall)
·
Delphi
(R.Gokieli)
·
L3 (M.Kienzle)
·
Opal
(S.O’Neale)
4.4 Future Fortran/CERNLIB support and long term
data access (J.Knobloch)
5. Update
on ongoing IT activities
5.1 Security
issues (D. Heagerty)
5.2 CLASP project update (D. Heagerty)
5.3 Castor and HPSS update (H. Renshall)
5.4 NAG C developer licences
(J.Knobloch)
5.5 User Revoking Policy (M.Delfino)
6. Actions outstanding
7. A.O.B. (includes raising of user issues not covered by agenda)
The chairman apologised for the change of data of the meeting. This was so as to include a full discussion of the LHC Computing Review in the context of FOCUS.
The following changes of FOCUS membership were
announced:
- Jacques Boucrot replaces Florence Ranjard as Aleph representative
- Manuel Delfino announced the appointment Jacques Altaber as the second deputy
IT Division leader, focusing on infrastructure and engineering computing. Pål
Anderson replaces him as IT/CS group leader and is therefore a new member of
FOCUS. Manuel showed an updated IT Division
organisational chart, highlighting the roles of the two deputy division
leaders.
The minutes had already been approved by E-mail. There were no further comments.
The chairman introduced the meeting by quoting a sentence from the last minutes: “FOCUS will oversee computing services at CERN, whether or not they are funded by CERN”. It has been pointed out that FOCUS was formed after the LEP experiments were already under way, i.e. when the funding for computing had already been resolved and the computing model defined.
LHC computing is significantly different, the funding model is different, CERN’s role has changed and we have the EU Data-Grid project. As a consequence, new management structures are being defined. The role of FOCUS has to be redefined in this context. The chairman sees two extreme possibilities:
· Leave all LHC computing to the new structures being developed, and concentrate on the rest of the programme
He hopes that the debate can begin today and no doubt the answer will be somewhere in between.
At this point the chairman has changed jobs and is no longer a particle physicist: he has been appointed Director of Computing at Oxford University. He will therefore have to stand down as FOCUS chairman at the end of his three year mandate, at the end of 2001. The secretary has also announced his wish to stand down at the same time. It would be difficult to find a new chairman and secretary before a new role for FOCUS has been agreed.
Another important item in the agenda is the question of LEP analysis. It had originally been foreseen to only address analysis until 2003 at this meeting. However, Jürgen Knobloch will also report on meetings that have taken place between Roger Cashmore, the LEP experiments and IT Division addressing the longer term issues.
Siggi presented a few slides outlining
the organisation and recommendations of the LHC Computing Review. These
recommendations are summarized in the report CERN/LHCC 2001-004. The review was
co-ordinated by a steering panel chaired by Siggi, whose members comprised the
chairs of the three specialised panels and the CERN Director for Scientific
Computing, with David Jacobs as secretary. In addition, representatives of IT
division and of the experiments were “in attendance”. The LHCC Chairman and the
CERN Director for Collider Programs were invited as observers.
The Worldwide Analysis / Computing Model
Panel (chaired by Denis Linglin):
·
Accepts the scale of resource requirements planned by the four
experiments
·
Recommends a distributed, hierarchical model as described by MONARC
·
Recommends the use of GRID technology. LHC computing is recognised as
the first large scale application of the GRID
·
Highlights the need for a well supported research network of 1.5-3 Gbps
per experiment,at affordable cost, by 2006.
The Software Project Panel (chaired by Matthias Kasemann):
·
Recommends joint efforts and common projects between experiments and
CERN-IT, and support for widely used products – in this context Siggi stressed
the need to clarify support for FLUKA and ROOT. – There are also many products
used only by single experiments; it is a task for future management committees
to try to improve this situation.
·
Recommends performing data challenges of increasing size and
complexity.
·
Encourages CERN to sponsor the transition to OO programming.
·
Identified many areas of concern, such as the limited maturity of
current planning and resource estimates, the insufficient development and
support of simulation packages, and the insufficient support and future evolution
of analysis tools.
The estimates derived by Management and Resources Panel (chaired by
Mario Calvetti) are subject to the following observations:
·
The current cost estimates are based on the forecast evolution of price
and performance of computer hardware, as set out in the PASTA reports. It seems
safe to extrapolate to 2006.
·
The hardware cost of the initial set-up of LHC distributed computer
centres (Tiers 0, 1 and 2) is estimated to be 240 MCHF, the CERN Tier 0+1
centre is about one third of the total. This figure is subject not only to the
PASTA extrapolation errors, but also to the significant uncertainties due to
the performance of the LHC, detectors, triggers, backgrounds etc.
·
The investment for the initial system should be spent in roughly equal
portions in the three years 2005, 2006 and 2007 (assuming LHC start-up in 2006
and design luminosity in 2007).
·
A major concern is that the core software teams are severely
understaffed. This is a problem specific to the experiments, that has to be
addressed by the participating institutes.
·
The planned reduction of CERN-IT staff is incompatible with CERN based
computing system and software support.
·
Maintenance of the LHC computing system from 2008 onwards, based on a
rolling replacement within a constant budget, requires about 1/3 of the
original investment worldwide per year (80 MCHF). This includes a steady
evolution in capacity as was already the case at LEP.
·
A common prototype should be set up as a joint project, reaching 50% of
the overall computing complexity (i.e. number of boxes, not CPU power) of one
LHC experiment by 2003/4 (estimated cost 18 MCHF, not included in initial
investment).
·
An agreement about construction and cost sharing of the prototype must
be set up now!
Finally, the review made the following general recommendations:
·
Set up an LHC Software and Computing Steering Committee (SC2), composed
of highest level software and computing management in the experiments, in
CERN-IT and in the regional centres, to steer deployment of the entire LHC hierarchical
system.
·
The SC2 establishes Technical Assessment Groups (TAGs) to prepare and
initiate tasks and projects
·
Each experiment must prepare a MOU for LHC Computing, describing the
funding and responsibilities for hardware and software, human resources etc..
Interim MOUs should be in place by end of 2001
Siggi concluded by saying that the LHC Computing Review is over, now
the funding of the LHC Computing Project is starting. The discussion was
delayed until after the following talk.
Federico summarised a recent meeting to
define the scope and goal of a Data Management TAG. It was decided to start
this group ahead of the creation of SC2 because the LHC Computing Review had
already identified Data Management as an item to be urgently addressed by a
TAG.
The goal of a TAG is to define the terms
of reference of new projects. The common understanding was that database
technology was not the focus of this first meeting. Rather it is important to
define areas of commonality, to assess requirements and technology and to focus
on short-term deliverables, in relation to the GRID and to the LHC prototype.
Coordination between experiments, IT, the prototype, the GRID etc. is of
paramount importance.
Actions resulting from the discussion
were that the experiments must work with IT to express their technical
requirements, that information must be collected on the non-common pieces, and
that database capabilities should be evaluated. Open questions are the
timescale for the first draft, and how to get there. It was decided to create a
small working group (one per experiment plus IT).
Discussion
The first question related to the scope
of the funding under discussion. Only offline computing is within the scope;
trigger and DAQ should already be included in the experiment budgets.
Hans Hoffmann then began to address the
question of how the Review relates to FOCUS. The Scientific Policy Committee
(SPC) and Committee of Council (CC) have been given copies of the Review
documents. A paper on building the LHC computing environment and on project
organisation is currently under discussion and will be presented to the June
meetings of the SPC, Finance Committee (FC), CC and Council, together with a
proposal to exercise the MONARC model “for real” over the next three years; a
proposal for the final system will be prepared at the end of 2003.
It is foreseen to have a project overview
board (executive board reporting to the DG and
advised on project requirements and goals by the SC2) will be set up
before the June Council meetings. One of the main aims for June is to secure
funding, not only for the structure outside CERN but also additional funding
for the CERN part of the project.
Of course the above structures are
exclusively aimed at LHC computing. The SC2 only addresses LHC: it will take on
the role of the LCB, but also talk of matters of concern to FOCUS. Since the
structure is not yet in place, Hans proposed to distribute to FOCUS, as soon as
it is available, the “green paper” to be presented in June, as a basis for
discussion at the next FOCUS meeting (ACTION). Preparatory discussions can take
place between Hans and the FOCUS chairman, but the experiments should also say
whether they need FOCUS and what they expect out of it (ACTION). The gut
feeling is that FOCUS will continue: the SC2 is a project follow-up structure
that will stay in place 2-3 years. FOCUS may suffer during this time but would
then take up again all its relevance. As an aside, the new structures are also
likely to step on the toes of HEPCCC.
Norman McCubbin understands the role of
the new structures as a stepping-stone, which will not be dismantled in 2004.
He expects them (or something similar) to remain in place until the project has
“reached design luminosity”.
Manuel Delfino believes that the
combination of FOCUS (to look in detail at requests) and COCOTIME (to allocate
resources) is very good, and essential for continuing IT operations (LEP
analysis, fixed target program, LHC test beams etc.). FOCUS is highly
appreciated for its guidance.
On the specific question of the testbed,
it was agreed that FOCUS could continue to review its operations, but should
not get involved in planning milestones over a period of years.
The question was asked why the Data
Management TAG was put in place before SC2 even exists. Federico and Manuel
replied that both the Review and the experiments had identified this particular
TAG as being needed urgently. It would be artificial to wait for the SC2 to set
it up officially, though of course SC2 may wish to modify it.
Paul asked whether this TAG would address
the interface to the EU Data Grid. There were conflicting opinions: Federico
felt that the prototype will be GRID-enabled, so the TAG has some role. Manuel
disagrees because TAGs are envisaged to be short-lived entities, so they cannot
be expected to take a coordination role – this is the role of SC2. Nevertheless
Paul expressed concern that the EU Data Grid is not mentioned explicitly in the
Review.
Finally it was made clear that R&D
projects which were being overseen by the LCB committee will be wound down, as
recommended by the review.
FOCUS thanks Siggi for
a clear overview and welcomes the structures being put in place to manage LHC
computing, both technically and managerially, and the well-defined conclusions
of the LHC Computing Review.
The new LHC structure
will be presented to committees in June, and we await feedback from the
committees. In the mean time we seek input to the FOCUS chairman by E-mail (Paul.Jeffreys@oucs.ox.ac.uk),
especially from the experiments (ACTION). This will be reviewed at the next
FOCUS meeting, in which we will look to define a new role for FOCUS (ACTION).
It was agreed to
consider the LHC prototype (previously called testbed) at the FOCUS meeting as
planned, as it is foreseen to have wider use than just the LHC experiments.
FOCUS supports the
early formation of the Data Management TAG, as proposed by the LHC Computing
Review.
Maya presented a first collection of
requirements for CVS and build services, obtained from discussions with individuals
in the LHC experiments and in IT/API
CVS servers require a local file system
backed up to a shared file system. For reliability and security reasons, the
files should be accessible only via the server, with login access reserved to a
few individual maintainers. Access control to the repository at user, group,
project and experiment levels is essential. The server needs sufficient disk
and memory to cope with large checkouts and with many simultaneous small disk
operations. Companion services are WWW access e.g. for browsing and for
scanning dependencies; a query manager; monitoring, quality control and metric
services; multi-repository services; automated documentation; integration with
the problem tracking system.
Build services require a reference
machine for each supported architecture, with well-defined and stable OS,
compiler, libraries. Sufficient local disk is needed for regular (nightly)
builds; access to the shared file system is needed to install the built
software. It should be easy to replicate the servers at regional centres.
Companion services are WWW services for automated/remote build management;
automated/remote build submission at various granularities, without interactive
login, and with E-mail notification; build analysis for monitoring problems
during builds; configuration management services; distribution services;
multi-site operations.
By way of conclusion, Maya observes that
all experiments already have specific facilities in place and asks herself
whether the differences are too great for a common approach. Probably an effort
should be made to identify commonalities to allow sharing of information and
experience on specific projects, to promote common maintenance and support
schemes, and help formulate informed recommendations.
The discussion tried to establish whether
a centrally provided solution is the way to go. Managed services such as
SourceForge already exist, so why haven’t the experiments already investigated
this possibility? The conclusion was that, if the experiment can agree amongst
themselves on what such a service should be, then IT division could consider
providing. But the initiative has to come from the experiments, with solid
requirements and a commitment to use the service provided.
A different question was whether IT
Division has any problem providing servers in a “non-standard” hardware
configuration as described in Maya’s presentation. This was not thought to be a
problem, and has in fact already happened with a machine requested in last
year’s COCOTIME round.
FOCUS thanks Maya for
her efforts to collect requirements for CVS and build services. Experiments
need to decide whether they would use a central secure CVS server, and if so
make a request for such a service (ACTION).
Harry presented a slide on behalf of
Jacques Boucrot. The migration of Aleph code to Linux is completed, with the
exception of two minor event generators that will be ported soon. Aleph uses
heavily the CERN Linux disk servers and CPU farms (both public and private) and
plans to decommission its Digital Unix SHIFT service in August. This requires
user scratch areas in AFS (under way) and the migration of working group disks
(several possible solutions under investigation).
For Delphi, the migration to Linux poses
no problems for the bulk of users, but there is still a lot of work for the
experts (the reconstruction is not yet commissioned, the LEP1 simulation has
not been ported, there are problems with sophisticated usage of the geometry
and calibration database). The event display is OK, but there are worries about
the status of GPHIGS.
The migration from HPSS is under way, but
can be painful and often requires expert help. An alternative to FATMEN is
being worked on; this is not too difficult since Delphi uses few of the
features of FATMEN. The most important data (real and simulated) will be
archived in Castor.
Conclusion: “We feel we have a right for
being optimistic!”
L3 has finished data production, but
needs to keep its SHIFT capacity until the end of 2003 for MC production. The
simulation and reconstruction programs work on Linux, as do most MC generators,
some of which do not work with g77 and need the Portland compiler. A problem is
the event display that works on HP-UX (using GPHIGS and Motif) but not on
Linux. Tests have been made to transfer data tapes to Castor, the FATMEN
catalogue has not been translated to a Castor system but no major problems are
foreseen.
In conclusion, L3 software works with
Linux and Castor, but the MC production has to stay on SHIFT due to lack of
manpower for the migration. The code will be frozen by end 2003, after which
access to the data is required in a transparent way.
The Opal migration to Linux is limited by
the available manpower. A thorough validation of the software on Linux (started
for the Y2K exercise) has been completed for the analysis and tracking
reconstruction, the physics comparisons are now acceptable. The validation of
the rest of the reconstruction and of the simulation is in hand, there are some
reservations about physics generators. Work on the graphics is in progress.
Steve presented several slides with
details of all the tools that have been checked on Linux, with no major
problem. Concerning data management, three proposals exist to use Castor below
FATMEN. For long term archiving the working hypothesis is to keep the existing
Linux/Castor system alive – no effort is available to embark on a migration to
OO or to a common LEP data format.
Harry asked why OPAL needs to keep its
SHIFT machines alive until 2003. There is in fact no strong reason, other than
the lack of manpower for completing the migration to Linux sooner.
As an aside, Tony Cass informed FOCUS
that IT division wishes to get rid of old data on 3480 tapes, which is taking
up a lot of space. There were no objections.
Jürgen addressed the topics of long-term
data storage, CERNLIB, and FORTRAN compilers on Linux.
Two meetings have taken place between the
LEP experiments and IT, chaired by Roger Cashmore. All experiments plan to
continue analysis as normal until 2003, then tailing off. New physics may
require Monte Carlo and its reconstruction as a black box for much longer. The
future of some required external components, such as GHPIGS, has to be
investigated. Intel will provide IA32 binary compatibility in hardware on its
forthcoming IA64 processors, so programs should continue to run.
To satisfy these requirements, it is
proposed to:
·
Have all data in Castor (including
migration to an eventual successor)
·
Support CERNLIB until end 2003 as
previously agreed by FOCUS
·
Keep “as is” CERNLIB (or an agreed
subset), ZEBRA, GEANT 3.21 on Linux “for ever” but with no new code, no user
support and no port to IA64. Recompilations will be done for major new Linux
versions, but must be tested by the experiments. If the tests fail, the last
working frozen version will be kept – this would be the start of a “museum”
system.
·
PAW support may be discontinued after
2003 if an appropriate replacement product exists.
The CERNLIB build procedure will be
further streamlined and documented for the last release in 2002, to allow
non-experts to continue building of CERNLIB on Linux.
A 2001 release of CERNLIB was not
planned, but is required to make the Castor client available. There is also
some repackaging and bug fixes. It is planned to make the current “NEW” version
“PRO” by early June. The different user communities should verify that this
does not cause a problem.
Concerning the LEP request for better
FORTRAN debugging tools on Linux, several options have been studied (PGI, NAG,
Lahey/Fujitsu). The first two have drawbacks, the last has not yet been studied
in detail. There is no conclusion yet.
Manuel Delfino concluded that there is no
need to re-discuss the Fortran and LEP long term analysis issue at the next
meeting, since agreement has been reached. Experiments should feel free to
bring any unforeseen problem to the attention FOCUS if necessary. Any offers to
shut down RISC machines ahead of the planned dates will be appreciated!
FOCUS would like to
thank all the contributors to the discussion concerning future LEP analysis. It
is reassured that the LEP experiments are making good progress in their
migration to Linux.
FOCUS is pleased to
note that the two meetings with Roger Cashmore have led to convergence
concerning the long term retention and use of LEP data, and notes the ‘pro’
release of CERNLIB foreseen for June 2001. Experiments should bring to the
attention of FOCUS any concerns that may develop regarding LEP analyses.
The Web security scan went smoothly.
Informational mails were sent to all owners of web servers, some of who felt
the mail was unclear. Recommended fixes were sent out in April, and replies
requested. It is proposed to repeat the scan in regularly, the next one being
in June. This was AGREED by FOCUS.
Denise also presented a summary of all
security incidents that occurred in 2000, the bulk of which was due to
compromised passwords. One worry is unauthorised use of FTP servers, it is
planned to look for vulnerable servers in the June scan.
It was suggested that Denise should make
regular security reports to FOCUS one a year. This was agreed (ACTION).
The goal of the CLASP project was to
reduce the number of login/passwords needed by users to access services they
are authorised to use. Feasibility studies were made which recommended the use
of Kerberos v5 technology, which was then tested on Linux, W2000 and the Grid
environment. However, worries were raised over the long term support for this
solution. This, coupled with the current migration to W2K and the uncertain
future of AFS, has lead to a new proposal for CLASP phase 2:
·
To design and pilot a password
synchronisation tool, including at least the Windows, AFS, Mail and AIS
passwords. Use of the tool would be optional.
·
To recommend off-site access mechanisms.
·
To design and pilot a tool for common
access control of web pages and files (“e-groups”) based on existing CERN
databases.
The general feeling from the audience was
that password synchronisation is not a high priority if it implies a lot of
work, the real need being single login as originally proposed. Password
synchronisation might help users who are confused by the different accounts
they have access to, in particular their mail and AFS accounts. Denise agreed
that the mail password is indeed a specific problem that comes up often and
that could be addressed within the scope of the original project.
Harry presented a proposal to terminate
the HPSS service by the end of 2001. HPSS does not support the Linux EIDE disk
servers, and there is no evidence coming from IBM that this will ever be more
than a niche product. On the other hand Castor is now sufficiently mature to
take over the user functionality previously done by HPSS.
IT division proposes to handle centrally
the migration of the 20TB of user files and 20TB of experiment files except
when experiments/users want to do it themselves. The migration procedure is
transparent for those users that access HPSS via the hsm command. Users who
explicitly use the /hpss file path will have to modify this to /castor after
their directories have been migrated. The data will be kept in HPSS after
migration (until final closedown at end 2001) but will be made inaccessible.
FOCUS agreed to the proposal and agrees
that the migration can start on 1st June.
The NAG C mathematical libraries are
replacing the CERNLIB math functionality for C++ developments. CERN and other
HEP institutes have site licenses, but licenses for individual developers at
small institutes would be expensive. Therefore CERN has negotiated an agreement
with NAG to distribute individual developer licenses. This was accepted by
Finance Committee, and 100 licenses are available – after signing the form
which can be found at http://cern.ch/Anaphe.
The license key will be provided by NAG to the individual, and will be valid
for a year, after which it can either be extended or transferred to another
user. This authorises a specific developer to use a single version of the
library (i.e. two licenses are needed if using two platforms). Sites with many
developers should purchase a site license – the break-even point is 5-8
licenses. Currently CERN does not charge LHC experiments for the individual
licenses, but the sharing of costs will be defined in the MoUs for computing.
The request by ACCU that people no longer
registered as CERN users keep their accounts cannot be accepted for
legal/liability reasons. The new leaders at HR (Vince) and Users Office (Chris)
are reviewing the whole situation of registering users and how to eliminate the
need for physical presence at CERN and will take the issue of computer usage
into account. ACCU and other relevant committees will be briefed whenever a
draft policy is elaborated. For the time being, IT will continue as today, only
closing accounts under extreme circumstances and after substantial personal
efforts to contact the user or people who should know the user.
Minuted/Section
|
Action
|
Who
|
Status
|
|
02/12/1999 16/4 |
Determine policy
for future use of FORTRAN CERN libraries |
IT Division, FOCUS |
See these
minutes, section 4.4 CLOSED |
|
02/03/2000-17/2 |
Report on LHC
Computing Review at next meeting |
D.Jacobs |
See these
minutes, section 2 CLOSED |
|
02/03/2000-17/4 07/12/2000-20/3 |
Propose to FOCUS
date for freezing of NICE95/NT |
F.Hemmer |
Freezing date to
be agreed at w2kmtf. Planned for FOCUS 24 |
|
02/03/2000-17/5 |
Definition of
LEP long term analysis strategy. |
R.Cashmore H.Hoffmann |
See these
minutes, section 4 |
|
08/06/2000-18/3 |
Organise
presentation on Linux migration at a future meeting |
M.Cattaneo, P.Jeffreys |
Planned for
FOCUS 22 |
|
08/06/2000-18/4.2 |
Identify full
set of software to be maintained in frozen operating system |
IT with
experiments. |
See these
minutes, section 4 CLOSED |
|
08/06/2000-18/5 |
Refine questions
on experiments' needs for remote backup and archive, including needs for any
features unique to ADSM |
IT Division |
Planned for
FOCUS 22 |
|
08/06/2000-18/6 |
Request
experiments to specify requirements for large shared LHC computing test-bed |
M.Cattaneo, P.Jeffreys |
Action to be
reviewed in the light of new role for FOCUS. Testbed review planned for FOCUS
23 |
|
08/06/2000-18/7.2 03/05/2001-21/5.5 |
Finalise with
FOCUS details of User Revoking Policy, following discussion at ACCU |
M.Delfino |
See these
minutes, section 5.5 CLOSED |
|
07/12/2000-20/6.3 |
Define IT/API
position concerning LHCb request to avoid code constructs known to cause
problems to the Visual C++ compiler |
J.Knöbloch |
Policy defined
and agreed with LHCb. CLOSED |
|
07/12/2000-20/6.3 |
Propose solution
to Fortran compiler problems on Linux |
J.Knöbloch |
See these
minutes, section 4.4 CLOSED |
|
07/12/2000-20/6.3 |
Arrange FOCUS
discussion on build servers |
M.Cattaneo, P.Jeffreys |
See these minutes, section 3 CLOSED |
|
07/12/2000-20/6.4 |
Explore with the
LHC experiments the possibility of a SUN batch service outside CERN |
H.Hoffmann |
See Memorandum
below. CLOSED |
Concerning the last action, Hans Hoffman presented a draft memorandum to the experiment
spokespersons, in which he states:
“With
the very tight resources for the LHC computing CERN does not see the
possibility to provide a 10% Production Service on Sun platform. As we are planning a "transparent"
distributed computing facility around the world, I would hope that you can
identify such a services elsewhere.
Nevertheless, CERN will supply the described
"SUNDEV" services permitting development work as discussed in the
FOCUS Committee.”
The was no other business
Minuted/Section
|
Action
|
Who
|
Status
|
|
08/06/2000-18/3 |
Organise
presentation on Linux migration at a future meeting |
M.Cattaneo, P.Jeffreys |
Planned for
FOCUS 22 |
|
08/06/2000-18/5 |
Refine questions
on experiments' needs for remote backup and archive, including needs for any
features unique to ADSM |
IT Division |
Planned for
FOCUS 22 |
|
08/06/2000-18/6 |
Request
experiments to specify requirements for large shared LHC computing test-bed |
M.Cattaneo, P.Jeffreys |
Action to be
reviewed in the light of new role for FOCUS. Testbed review planned for FOCUS
23 |
|
03/05/2001-21/2 |
Distribute to
FOCUS the Council papers defining new structures for LHC computing |
H.Hoffmann |
|
|
03/05/2001-21/2 |
Send comments
concerning future role of FOCUS to chairman, by E-mail |
H.Hoffmann,Experiments |
|
|
03/05/2001-21/2 |
Organise
discussion on future role of FOCUS |
Secretary, Chairman |
Planned for
FOCUS 22 |
|
03/05/2001-21/3 |
Decide on need
for a central CVS server |
Experiments |
|
|
03/05/2001-21/5.1 |
Plan annual Security
report at FOCUS 25 |
Secretary |
|
