Bild meeting 30.09.2011 CERN: Joel, Mario, Federico, Zoltan, Vlad, Krzis, Teodor, Elisa Phone: Andrei Dirac v6 status: Andrei: almost ready, a prerelease ready and tested on two systems at Marseille and Lyon will be installed on volhcb13 the instance in Lyon, called pre-6, is in production for the French NGI Joel: asked for clear instructions to move from serie v5 to v6 Andrei: the v5r14 release is a fake release to make possible the transition, it has to be installed ONLY to install the new installation tools, and right after v6 can be installed The above said, for installation with cli. Joel asks also about installation scripts: in this case no need to pass through the v5r14 Andrei will provide instructions for both installation procedures, and then Joel will decide which way to go Tests to be performed on the volhcb13, with DIrac v6: services used specifically by LHCb, like LFC, direct CREAM submission, and submission through WMS, Storage Element if problem are spotted, the v6r0 release will be patched. Fede: propose that Diracv6 could be installed on volhcb12, and people who usually work on it will spot possible bugs. Plan: -wait for the first Dirac v6 prerelease -wait for instructions for the installation -install on volhcb12 the v6 prerelease -if no problem, first LHCb release based on Dirac v6r0 RSS: RSS almost ready to store the SE status, instead than the CS. When a client request the SE status, if not present in the RSS, the status will be looked up in the CS (so during a period the status will be stored in both, RSS and CS) When RSS will go in production, there will be no active policy (active policies would automatically change the SE status. This will happen only later) New TokenAgent: a token is a temporary ownership, e.g. if Vlad bans a SE, he will be assigned a token. At the token expiration, the owner can decide whether to renew or not. Andrei: RSS will be included in Diracv6r1 it is only used into ReplicaManager, where the SE status is asked to the RSS. this has been reverted back in v6r0: in v6 ReplicaManager asks SE status to CS. Some worries about the timeline to have it released. Andrei suggests this strategy (i.e. incremental strategy): release a version of LHCBDirac with NO new feature, on top of a new Dirac release, so that certification should be easy. Then, new LHCBDirac release , based on the same Dirac, with the new LHCb features. In this way DIRAC and LHCbDirac new features are decoupled and certification should be easier. Conclusion: Rss will be included in v6r1, and ReplicaManager will use it Also other components can use the RSS: scripts to ban/unban SE, Transformation system, and few other LHCBDirac releases: - no need for a patch today - based on what will be said at tomorrow's core meeting, we'll decide what to put in the LHCbDIRAC release that will be used at the start of the reprocessing. We'll first prepare an LHCbDIRAC release based on the DIRAC v5r13p22, that will be certified, and that will hopefully be very short lived, since a new release based on DIRAC v6r0 will be made as soon as possible including also these changes. ---- Reports from the groups Framework: major upgrade of the framework will happen with v6 Then, the changes in the Agents to use the new features of the framework, will come in a later release (not in v6r0). ---- WMS few minor patches, in v5r14. no major development. JObWrapper: considerations about the running status of the job: now the job status is 'running' since the input data resolution and sometimes the IO takes long time so the job is declared stalled. It might be considered to set the job status as 'running' only after the initial phase of input data resolution To be considered. -------DMS Krzis: issued pull request for some FTS monitoring improvements Elisa: Resources/StorageElement included in Dirac v6r0, this include the 'check' Access status for ARCHIVE SEs. SEUsageAgent: new implementation of the agent login and the input file parser to take into account new space tokens schema. Committed and under testing in volchb12, but not tagged yet (before I need to test it with complete storage dumps, that sites still don't provide) StorageHistoryAgent: serious performance problems, trying to improve the execution time of the agent. Currently is 16-17h to browse the whole BKK tree . Ongoing. -------RSS Vincent: migrated to RSS some scripts from Roberto's area and included in RSS agents. The gfal api to get the space usage from SRM is broken in last lcg bundle. Still waiting for a fixed version from gfal developers. ----- production/transformation Fede: several minor fixes --------BKK Zoltan: implementing new feature for the web interface Added the data quality and TCK. After, the functionality of web interface and cli will be equivalent Oracle update: new xml catalog implemented, the new format is not used now in production and it will be used in the reprocessing. It should be included in a new LHCB release before the reprocessing starts. A parenthesis about the plans for next releases: the target is to have a LHCBDirac release based on Diracv6 for the start of the reprocessing, and if this is not achieved we should have ready a fallback solution, based on Dirac v5r13. ---- web portal ntr, Matvey on vacation --------- AOB ------- Python Joel: most of agents and services are running with Python2.6, but installing them with this version of Python, before summer,was not easy. With the dirac-install there is a switch to decide which python version to use, with the cli no. Andrei: for setting the python version with the cli, the dirac.cfg has a session for the local installation where the version of pyhon can be set. Agreed to use python 2.6 for all voboxes ---- Re-installation of volhcb12: a notice will be given before re-installing, and the machine will not be re-installed before receveing green light from the developers that usually work with it