Presentation is loading. Please wait.

Presentation is loading. Please wait.

GridPP CM, ICL 16 September 2002 Roger Jones. RWL Jones, Lancaster University EDG Integration  EDG decision to put short-term focus of effort on making.

Similar presentations


Presentation on theme: "GridPP CM, ICL 16 September 2002 Roger Jones. RWL Jones, Lancaster University EDG Integration  EDG decision to put short-term focus of effort on making."— Presentation transcript:

1 GridPP CM, ICL 16 September 2002 Roger Jones

2 RWL Jones, Lancaster University EDG Integration  EDG decision to put short-term focus of effort on making the ATLAS DC1 production work on 1.2 and after  Good input from the EDG side. Effort from ATLAS, especially UK, Italy and CERN (Notably Frederic Brochu, but also others: Stan Thompson, RJ, and Alvin Tam joining now)  The submission seems to work `after a fashion’ using multiple UIs and one production site  More problems using multiple production sites, esp. data replication  Inability to access Castor using Grid tools was a major problem, fixed  Large files sizes also a problem  Interfaces to catalogues need work  Encouraging effort, expect full production quality service for demonstrations in November  Note: analysis requires more work – the code is not already `out there’  Integration with other ATLAS Grid efforts needed  Evaluation session on Thursday at RHUL Software Week

3 RWL Jones, Lancaster University Exercises  Common ATLAS Environment pre-installed on sites  Exercise 1: simulation jobs  50 jobs, 5000 events each, 5 submitters  One-by-one submission because RB problems  Running restricted to CERN  1 job 25h real time, 1Gb output  Output stored on the CERN SE  Exercise 2 with patched RB and massive submission  250 shorter jobs (several hours, 100Mb output)  CERN, CNAF, CCIN2P3, Karlsrurhe, NIKEF, RAL  One single user  Output back to CERN – SE space is an issue  Also have `fast simulation in a box’ running on the EDG testbed, good exercise for analysis/non-installed tasks

4 RWL Jones, Lancaster University Issues  There is no defined `system software’  The system managers cannot dictate the shells to use, compilers etc  Multiple users will lead to multiple copies of the same tools unless a system advertising what is installed and where is available (PACMAN does this for instance)  A user service requires binaries to be `installed’ on the remote sites – trust has to work both ways

5 RWL Jones, Lancaster University Software Away from CERN  Several cases:  Software copy for developers at remote sites  Software (binary) installation for Grid Productions  Software (binary) download for Grid jobs  Software (source) download for developers on the Grid  Initially rely on by-hand packaging and installation  True Grid use requires automation to be scalable  Task decomposes three requirements  Relocatable code and environment  Packaging of the above  Deployment tool (something more than human+ftp!)

6 RWL Jones, Lancaster University World Relocatable Code  Six months ago, ATLAS code was far from deployable  Must be able to work with several cases:  afs for installation  No afs available  afs present but not to be used for the installation because of speed (commonplace!)  Significant improvement in this area, with help from John Couchman, John Kennedy, Mike Gardner and others  Big effort in reduction of package dependencies – work with US colleagues  However, it takes a long time for this knowledge to become the default – central procedures being improved (Steve O’Neale)  For the non-afs installation the cvsupd mechanism seems to work generally, but patches are need for e.g. Motif problems  Appropriate for developers at institutes, but not a good Grid solution in the long-term

7 RWL Jones, Lancaster University An Aside: Installation Method Evaluation  Previous work from Lancaster, now big effort from John Kennedy  Cvsupd  Problems with pserver/kserver; makefiles needed editing  Download took one night  Problems with CMT paths on the CERN side?  Problems fixing afs paths into local paths – previously developed script does not catch all  NorduGrid rpms  Work `after a fashion’. Does not mirror CERN, much fixing by hand  Much editing of makefiles to do anything real

8 RWL Jones, Lancaster University An Aside: Installation Method Evaluation (II)  Rsynch method  Works reasonably at Lancaster  Hard to be selective – easy to have huge downloads taking more than a day in the first instance  Only hurts the first time – better for updates  Official DC1 rpms  Circularity in dependencies cause problems with zsh  Requires root privilege for installation

9 RWL Jones, Lancaster University Installation Tools  To use the Grid, deployable software must be deployed on the Grid fabrics, and the deployable run- time environment established  Installable code and run-time environment/configuration  Both ATLAS and LHCb use CMT for the software management and environment configuration  CMT knows the package interdependencies and external dependencies  this is the obvious tool to prepare the deployable code and to `expose’ the dependencies to the deployment tool (MG testing)  Grid aware tool to deploy the above  PACMAN is a candidate which seems fairly easy to interface with CMT, see following talk

10 RWL Jones, Lancaster University CMT and deployable code  Christian Arnault and Charles Loomis have a beta-release of CMT that will produce package rpms, which is a large step along the way  Still need to clean-up site dependencies  Need to make the package dependencies explicit  Rpm requires root to install in the system database (but not for a private installation)  Developer and binary installations being prepared, probably also needs binary+headers+source for a single package  Converter making PACMAN cache files from auto- generated rpms seems to work

11 RWL Jones, Lancaster University The way forward?  An ATLAS Group exists for deployment tools (RWLJ convening)  LCG have `decided’ to use SCRAM.  Grounds for the decision seemed narrow, with little thought to implications outside of LCG  If this is to be followed generally, should reconsider strategy  What about using SlashGrid to create a virtual file system?


Download ppt "GridPP CM, ICL 16 September 2002 Roger Jones. RWL Jones, Lancaster University EDG Integration  EDG decision to put short-term focus of effort on making."

Similar presentations


Ads by Google