CERN Regional Operations Center (ROC) Report
============================================
* Grid site UIOWA-LCG2 (CE: lcg-ce.its.uiowa.edu) requested to be permanently retired from the EGEE grid. The site's grid services have been removed from the GOC database and the site has been marked as being in scheduled downtime. CERN ROC changed the status of the site in GOC DB to "Closed". Consequences of the 'unregistering' of the sites on monitoring tools still unclear. Removal process from WLCG still undefined in details. Working on clarifying this with central grid operations team.
 
 


CERN Tier-0 Site Report (2 April 07)
====================================

CASTOR2
-------
* New version of the SRM v11 is being tested. The new release fixes some problems reported by LHCb related with the TURLs (Transfer URLs) returned for ROOT applications and CERN and PIC
* preparing the new CASTOR2 stager for the ATLAS T0 tests
* preparing Castor nameserver DB move on Monday (2/4)
* Provided test machines for the experiments to test new gcc & libs, email sent to info-experiment If no showstopper, deployment next Wednesday
* Network switch exchange will go ahead as planned next Monday (IP 301, 302, 303 in 513-C). Batch will be stopped for that.
* New kernel deployment on SLC4.
* Ongoing work on deploying/testing the new LSF plugin.
* An important bug was discovered on the 2.1.2-6 release that led to a crash of the stager in case a prepareToGet was done with several files. This bug has been fixed in the meantime.
* SRMv22 is running stable (modulo few fluctuating problems which have been understood and fixed in CVS a cumulative bugfix release will be produced mid next week). However, initial stress tests have shown problems which need to be further analyzed by IT/GD (Flavia).
* XROOTD has now been installed on the preproduction setup and is being tested by Alice. First tests revealed some configuration issues that are being solved in collaboration with the XROOT team.
* RAL are officially asking for Disk1Tape0 support (see Savannah #23303).


AFS
---
* afs96 (serving ATLAS and LHCB, LHCB complained) is overloaded, but the exact cause isn't understood yet = still under investigation, some volumes will have to be moved to (more performant) mirror disks.
* AFS "full disk restore" has been implemented.
 

Physics Database Service
------------------------
* We have started the tests of a quad-core server and compared it against current RAC hardware. Tests have been performed using Oracle-generated workloads. RHEL4 64 bit and Oracle 64 bit are used for the test. First results show an increase in performance of about a factor 5 in favour of the quad-core, i.e. 1 quad core performs as a 5-node RAC. Further tests are in progress using physics applications. Preliminary results show a consistent gain in performance for Oracle Streams and COOL.