Priorityuser .

Description

Our goal here is allocate 25% of our pledge resources to users with priorityuser voms role. The first part is make this users authenticate to a special user in our farm, using gums. First editing our $VDT_LOCATION/tomcat/v55/webapps/gums/WEB-INF/config/gums.config like that

%CODE{html}% <vomsUserGroup name='priorityuser' access='read self' description='' vomsServer='cms' matchFQAN='exact' acceptProxyWithoutFQAN='false' voGroup='/cms' role='priorityuser'/>

<groupAccountMapper name='priorityuser' description='' accountName='cmsprio'/>

<groupToAccountMapping name='priorityuser' description='' accountingVoSubgroup='uscms' accountingVo='CMS' userGroups='priorityuser' accountMappers='priorityuser'/> %ENDCODE%

and add in the end of this file, after = groupToAccountMappings= , in some place into your mappings the entry priorityuser It will be necessary restart your tomcat and mysql,

 /etc/init.d/tomcat-55 stop; /etc/init.d/mysql stop;  /etc/init.d/mysql start; /etc/init.d/tomcat-55 start

and don't forget to add cmsprio as your common user:

adduser  cmsprio 

Just if you use NIS,

cd /var/yp
make

Make sure that you can log in into your worker nodes as this user.

The second part is make your condor jobmanager assign, based in the user name, a dedidated group to its jobs: %CODE{perl}% vim $VDT_LOCATION/globus/lib/perl/Globus/GRAM/JobManager/condor.pm map { if ($_->[0] eq "LOGNAME") {$logname = $_->[1]; } } @environment; if ($logname =~ m/cmsprio./) { $AccountingGroup = "group_cmsprio." . $logname; } %ENDCODE% and adding it, in the same file above: %CODE{perl}% print SCRIPT_FILE "+AccountingGroup = \"$AccountingGroup\"\n"; %ENDCODE% After that, we have to define in our condor_negotiator machine (in our case, our gatekeeper) these groups and quotas:

vim /scratch/condor/condor_config.local
GROUP_NAMES = group_cmsprio
GROUP_QUOTA_group_cmsprio = 50
GROUP_AUTOREGROUP_group_cmsprio = True

and make condor re-read its configuration file, using : condor_reconfig In this case we have dedicated 50 job slots (25% of our 200 job slots) to cmsprio user.

About the storage area, we created it in our dCache admin the correspondent directory

mkdir /pnfs/sprace.org.br/data/cms/store/user/priorityuser

and edit our /etc/grid-security/storage-authzdb in our gridftp servers to grant permission to cmsprio write/read it:

authorize cmsprio read-write 0 110 111 / /pnfs/sprace.org.br/data/store/user/priorityuser /pnfs/sprace.org.br/data/

The same must be done to /store/group and /store/results