Job Robots Troubleshooting

Description

All jobRobots are aborted in our farm, looking at the page http://jobrobot.web.cern.ch/JobRobot/aborted_081019.html#T2_BR_SPRACE

 BrokerHelper: no compatible resources
 request expired

First we checked some corruption in our CMSSW installation, running a crab using the same version of CMSSW pointed in http://jobrobot.web.cern.ch/JobRobot/summary_081019.html following instructions at http://www.sprace.org.br/Twiki/bin/view/Main/EntryDescriptionNo53

May be this error is relatade with an ambiguous BDII publication due a requirement that makes the matchmaking to fail, actually

Member("osg-se.sprace.org.br",other.GlueCESEBindGroupSEUniqueID) 

In our BDII was:

objectClass: GlueSchemaVersion
GlueCESEBindGroupCEUniqueID: osg-ce.sprace.org.br:2119/jobmanager-condor-cms
GlueCESEBindGroupSEUniqueID: osg-se.sprace.org.br
GlueCESEBindGroupSEUniqueID: osg-se.sprace.org.br
GlueSchemaVersionMajor: 1 

Note that GlueCESEBindGroupSEUniqueID: osg-se.sprace.org.br appears twice. To remove it, we need to fix GIP that collects information to CEMon. Changing directly the file

/OSG/gip/var/ldif/osg-info-static-cesebind.ldif

seems that it doesn't work. So, we changed the gip-attributes that is read by configure_gip to make this file

vim /OSG/monitoring/gip-attributes.conf
OSG_GIP_DISK="0"
/OSG/vdt/setup/configure_gip

And it can be checked to work with

ldapsearch -x -LLL -p 2170 -h lcg-bdii.cern.ch -b mds-vo-name=SPRACE,mds-vo-name=local,o=grid > jobrobot.txt

Updates

Fulano em dd/mm/aaaa

Coloca o que fez.

Ciclano em dd/mm/aaaa

Mais comentarios

-- MarcoAndreFerreiraDias - 19 Oct 2008