Choose timezone
Your profile timezone:
PASTE meeting
Description
CERN, bat 2 R-030
To join the telephone conference, click here: https://audioconf.cern.ch/call/0165202
To dial in to the conference:
a. Dial +41 22 7676000
b. Enter access code 0165202
Production and PASTE Meeting, 30th September 2008, CERN
CERN: J.Closier, P.Charpentier, A.C.Smith, R.Nandakumar,Z.Mathe,V.Romanovskiy
Phone: M.Sapunov, N.Brook, R.Graciani
Chair/Minutes: A.Smith
0. Existing Action List
(07) - 27/11/07 - Zoltan, Elisa - Ongoing new bookkeeping system development
(57) - 24/06/08 - Vladimir,Roberto - Automatic publishing of BDII / GOCDB information in the CS using
GOCDB site names and provide a summary of recent changes
(70) - 12/08/08 - Joel,Ricardo - Perform and document DIRAC3 installation in the LHCb release area
at CERN with accessible CS file (Ricardo to update the install script for LCG tag)
(79) - 26/08/08 - Vladimir - Update CS to use new host aliases
(82) - 01/09/08 - Ricardo/Adri - to provide information from accounting to allow GridMap for LHCb VO T1 activity
(83) - 01/09/08 - Stuart,Roberto - To perform glexec tests
(86) - 09/09/08 - Andrei - Document the job finalization output data upload policy
(90) - 09/09/08 - Joel - Use new ProductionOutputData JDL parameter in modules instead of creating LFNs
(92) - 16/09/08 - Ricardo - Add publishing of information to batch monitoring tool
(93) - 16/09/08 - Andrew - Improve RAW data monitoring tools
(94) - 23/09/08 - Roberto, Marcos - Using SystemLogging service for dynamic services status monitoring
(95) - 23/09/08 - Roberto - SLS and SAM sensors development
(96) - 23/09/08 - Roberto, Matvey - Dashboard SAM tests visualization, porting to DIRAC framework
(97) - 23/09/08 - Andrei - release and install DIRAC v2r21 with the new Stager code
(98) - 23/09/08 - Philippe - Check SE definitions in the DIRAC2 CS for conformance with SRM v2 end-points
(99) - 23/09/08 - Roberto - arrange for LFC entries change directly in the ORACLE back-end
(100) - 23/09/08 - Joel - ask to lower priorities of the LHCb LXBATCH non-grid queues
(101) - 23/09/08 - Zoltan - check memory consumption by the new BookkeepingManager processes.
(102) - 23/09/08 - Matvey, Adria - DIRAC Web portal installation on volhcb11, documenting the installation
procedure
(103) - 23/09/08 - Paul, Stuart - produce structured HTML version of the Shifter's guide
Update on ongoing actions:
(07) - Zoltan will begin the process of inserting the new processing pass.
(57) - Vladimir proposed the need to finalize the representation of site information in the CS.
(70) - Discussion of this action below spawned two further actions (closed)
(79) - Vladimir stated that there are inconsistencies in the twiki containing the aliases. He will sit with Joel to be sure which are correct. Action updated.
(83) - The delay of real data should increase the priority of this task since more time can now be devoted to it.
(92) - Ricardo reported the development is done and sending messages without errors in the client. Status update sent to the system admin (Danille Philippe) to check whether messages were correctly received. No response yet. (closed)
(97) - Done (closed)
(98) - Reassign to Raja to coordinate with Roberto (99) and update CS with SEs contained in the twiki,
(99) - Waiting until Oct 1st.
(100) - Joel lowered the 'production' share for non grid queues. (closed)
(101) - The source of the memory leak was discovered to be the agent transferring the messages between the bookkeeping servers. This is no longer used and has been stopped. No further memory leaks have been observed. (closed)
(102) - Matvey reported the installation in almost complete.
Updated action list:
(07) - 27/11/07 - Zoltan, Elisa - Ongoing new bookkeeping system development
(57) - 24/06/08 - Vladimir,Roberto - Automatic publishing of BDII / GOCDB information in the CS using GOCDB site names and provide a summary of recent changes
(79) - 26/08/08 - Vladimir/Joel - Check twiki contents and update CS to use new host aliases
(82) - 01/09/08 - Ricardo/Adri - to provide information from accounting to allow GridMap for LHCb VO T1 activity
(83) - 01/09/08 - Stuart,Roberto - To perform glexec tests
(86) - 09/09/08 - Andrei - Document the job finalization output data upload policy
(90) - 09/09/08 - Joel - Use new ProductionOutputData JDL parameter in modules instead of creating LFNs
(93) - 16/09/08 - Andrew - Improve RAW data monitoring tools
(94) - 23/09/08 - Roberto, Marcos - Using SystemLogging service for dynamic services status monitoring
(95) - 23/09/08 - Roberto - SLS and SAM sensors development
(96) - 23/09/08 - Roberto, Matvey - Dashboard SAM tests visualization, porting to DIRAC framework
(98) - 23/09/08 - Philippe - Check SE definitions in the DIRAC2 CS for conformance with SRM v2 end-points
(99) - 23/09/08 - Roberto - arrange for LFC entries change directly in the ORACLE back-end
(102) - 23/09/08 - Matvey, Adria - DIRAC Web portal installation on volhcb11, documenting the installation procedure
(103) - 23/09/08 - Paul, Stuart - produce structured HTML version of the Shifter's guide
(104) - 30/09/08 - Developers - Read the file merging proposal from Philippe ahead of next PASTE meeting.
(105) - 30/09/08 - Joel - Lower user share on non-grid queues at CERN
(106) - 30/09/08 - Ricardo - Produce a list of what DIRACInstall is doing
(107) - Ricardo/Philippe/Joel/Hubert/MarcoC - Produce proposal for how to perform standardised installation using install_project/CMT.
1. Production (J.Closier)
Jobs submitted in the 3039 production for testing PIC were all submitted. Several jobs failed because the Ancestor depth of the production was not set resulting in the jobs to be incorrectly targeted to PIC. The remainder of the jobs completed the staging without incident and are now running.
Joel is testing the ParticleGun workflow on the LHCb-Development system. Matching of the jobs was failing due to intermittent problems with the voms-proxy-init with the proxy renewal server, this was resolved by Ricardo during the PASTE meeting.
The usage of disk space at the Tier1s was discussed and a proposal for the removal of files will be presented by Andrew.
2. DIRAC3 tasks
From Ricardo's Mail:
Jobs trying to run GaudiPython jobs with DIRAC3 were failing with the following error message:
Traceback (most recent call last):
File "/raid/expt-sw/SL4/lhcb/lib/lhcb/GAUDI/GAUDI_v20r1/InstallArea/scripts/gaudirun.py", line 42, in <module>
c.run()
File "/afs/cern.ch/sw/Gaudi/releases/GAUDI/GAUDI_v20r1/InstallArea/python/Gaudi/Main.py", line 53, in run
import GaudiPython
File "/afs/cern.ch/sw/Gaudi/releases/GAUDI/GAUDI_v20r1/InstallArea/python/GaudiPython/__init__.py", line 12, in <module>
from Bindings import *
File "/afs/cern.ch/sw/Gaudi/releases/GAUDI/GAUDI_v20r1/InstallArea/python/GaudiPython/Bindings.py", line 20, in <module>
import Pythonizations
File "/afs/cern.ch/sw/Gaudi/releases/GAUDI/GAUDI_v20r1/InstallArea/python/GaudiPython/Pythonizations.py", line 59, in <module>
gbl.AIDA.IHistogram1D.__repr__ = _printHisto1D
AttributeError: class _global_cpp has no attribute 'AIDA'
After some debugging it became clear that problem is related to the refex.rootmap files that are present on some of the library directories of the LHCB software distribution. In order to avoid problems with long LD_LIBRARY_PATH environmental variable, that we have seen in the past when running on arbitrary GRID sites. DIRAC is traverse all the directories in LD_LIBRARY_PATH and make links to all files from a single local directory, updating the LD_LIBRARY_PATH afterwards.
Since this reflex.rootmap files all have identical names, only the first one appearing remains.
For the moment the solution has been to comment the line doing this LD_LIBRARY_PATH compression. This is implemented on the HEAD version of DIRAC an a new release will have to be done to get it running in production.
The root refection mechanism will read any *.rootmap file present on any directory in the LD_LIBRARY_PATH variable. So other possible solutions are:
- give different names to this files (this can be done at the time they are created or when linking)
- they can be concatenated in a single file. Actually the current files are the concatenation of the rootmap files corresponding to each of the shared libraries present in a given directory.
Conclusion: This will be brought up at the core software meeting which Ricardo agreed to attend.
3. AOB
DIRAC installation and configuration (Philippe)
Two parallel procedures for installation and configuration in the LHCb computing project software with the same functionality.
Joel and Hubert installed DIRAC3 in the LHCb release area using standard tools for configuring the environment with the small problem of the vomses file/directory which was resolved. It appears that the version of the voms-proxy-init used in DIRAC is old and doesn't support the option to resolve the vomses file issue with an option (related to the ownership of the le/directory which must be the user or root which is a problem in the shared area).
Is it possible to have everything in a single deployment using the LCG applications area. A small modification to the CMT requirements file need to pick up the externals from the LCG-AA.
DIRAC can be distributed with any version of python that the lcg bindings are support. CMT_CONFIG to used to define config and subsequently obtain the name used within DIRAC. Also, need CMT to find out which version of python is used. Then can configure the lcg-aa and choose which version of the external packages.
Another problem was discovered when installing two platforms. Scripts changed at execute time to contain a preamble which gets confused whether it should execute 32 or 64 versions. Can make use of an environment variable to determine which python to use i.e. DIRACPYTHON.
In the script directory create a link to the destination script? Problem with ensuring that the environment is set correctly. Currently the environment is checked and verified using the DIRACEnvironment module which is imported at the same level as the scripts.
Conclusion: Should try and incorporate into standard LHCb tools while maintaining the ability to install DIRAC in a stand alone way.
There are minutes attached to this event.
Show them.