 |
 |
Hi Maarten, I am going to reconfigure our lcg-CE and BDII_site (on the same node) and trying to avoid the pitfalls mentioned below. I am going to use the latest YAIM (in 3.0.2 update 24) and it is the first time I run it since 3.0.1-x was released. I also need to enable a VO with DNS style name, which I understand requires a separate configuration file in the vo.d directory. 1) Can you confirm that BDII_CE_URL="ldap://$CE_HOST:2135/mds-vo- name=local,o=grid" still good for my lcg-CE? The release documentation is vague and seems to suggest otherwise 2) according to the YaimGuide301, queue management has changed, with the introduction of (QUEUE NAME)_GROUP_ENABLE variables. The variables VO_$VO_QUEUES no longer appear in the examples/site-info.def Vega's experience below seems to suggest the latter variables are still used instead. Should I stick with the old variables for the existing VOs and use the new queue management for the new VO only? cheers, Gianfranco On 26 Apr 2007, at 10:58, Vega Forneris wrote: > > Hi Maarten, > > yes, at last I was able to reconfigure correctly the lcg-CE. I > reset the correct BDII_CE value and had to reenable the variables > VO_$VO_QUEUE values which I had understood that were not needed > anymore. > > Cheers > Vega > > > > "Maarten Litmaath, CERN" <[log in to unmask]> > Sent by: LHC Computer Grid - Rollout <LCG- > [log in to unmask]> > 25/04/2007 22:19 > Please respond to > LHC Computer Grid - Rollout <[log in to unmask]> > > > To > [log in to unmask] > cc > Subject > Re: [LCG-ROLLOUT] - Production - lcg-CE - Update 21 > > > > > > On Thu, 19 Apr 2007, Vega Forneris wrote: > > > Hi *, > > > > I've just reconfigred my CE (lcg like), it works if targeted by > > globus-job-run (both fork and lcgpbs scheduler), but it > disappeard from > > top BDII. > > > > > > > > Looking in the specific release note, I found: > > > > > > In the site-info.def file is necessary to define the BDII_CE_URL > parameter > > as follows: > > > > BDII_CE_URL="ldap://$CE_HOST:2170/mds-vo-name=resource,o=grid" > > It says to do that when the site BDII is _not_ on the CE. > Furthermore, AFAIK it only applies to the gLite CE; > the LCG-CE still runs an MDS on port 2135 instead. > > > so I changed the value accordly (well...just cut and paste!) and > launched > > the reconfiguration of the bdii side: > > > > /opt/glite/yaim/bin/yaim -r -s /home/shared/conf/esrin-site- > info.def -n > > lcg-CE_torque -n BDII_site -f config_bdii > > > > but nothing changed...should I wait (maybe it needs to repopulate > the top > > BDII) or should I be worried? > > The repopulation is done about every 2 minutes... > > Apparently you fixed it somehow: your CE does appear now. >