General information on the condor installation

For the condor batch system installation we configure the gLite CE as a condor submitter machine which uses the machine CONDOR-MANAGER as condor master node.

The local condor deamons arte configured to run as root to not interfere with the user condor, which is configured by gLite.

Local Condor installation

We get the package condor-6.8.5-linux-x86-rhel3-dynamic-1.i386.rpm and istall it :

# rpm -ihv condor-6.8.5-linux-x86-rhel3-dynamic-1.i386.rpm 
 Preparing...                ########################################### [100%]
    1:condor                 ########################################### [100%]
We create a link to condor-local
 # ln -s /opt/condor-6.8.5 /opt/condor-local 
We change some variables in /opt/condor-6.8.5/etc/condor_config from
 HOSTALLOW_WRITE = YOU_MUST_CHANGE_THIS_INVALID_CONDOR_CONFIGURATION_VALUE
to
 HOSTALLOW_WRITE = *.yourdomain
(EXAMPLE: HOSTALLOW_WRITE = *.pic.es)
We configure condor as submitter machine with root as the owner and CONDOR-MANAGER your LRMS headnode.
 # export CONDOR_CONFIG=/opt/condor-local/etc/condor_config 
and configure:
 # /opt/condor-local/sbin/condor_configure   --owner 0 --central-manager=CONDOR-MANAGER --type=submit --install-dir=/opt/condor-local
 (EXAMPLE: /opt/condor-local/sbin/condor_configure   --owner 0 --central-manager=vwn02.pic.es --type=submit --install-dir=/opt/condor-local)
We edit line 60 condor_config in /etc/condor-local/etc/local.vce02.pic.es to
 CONDOR_IDS=0.0
and set the variable
 UID_DOMAIN = $(FULL_HOSTNAME)
 UID_DOMAIN = $(FULL_HOSTNAME)
We change the file permission
 
 # chmod 0.0 -R /opt/condor-local/
We copy the init file to /etc/init.d
 # cp /opt/condor-6.8.5/etc/examples/condor.boot /etc/init.d/
and modify the line 23 form
 MASTER=/usr/local/condor/sbin/condor_master
to
 MASTER=/opt/condor-6.8.5/sbin/condor_master 
and line 38 from
 pid=`$PS | grep condor_master | grep -v grep | awk '{print $2}'`
to
 pid=`$PS | grep root | grep condor_master | grep -v grep | awk '{print $2}'`
We start the service
 # /etc/init.d/condor.boot start
and have a look if everything is running fine
 # ps -aux | grep condor | grep root
 root     17472  0.0  1.0 11596 2836 ttyp0    T    12:13   0:00 emacs -nw /etc/init.d/condor.boot start
 root     32628  0.0  1.0  6788 2892 ?        S    12:35   0:00 /opt/condor-local/sbin/condor_master
 root     32629  0.0  1.3  8212 3524 ?        S    12:35   0:00 condor_schedd -f
 root      3034  0.0  0.2  3700  676 ttyp0    S    12:40   0:00 grep condor

 # /opt/condor-local/bin/condor_status
 
 Name          OpSys       Arch   State      Activity   LoadAv Mem   ActvtyTime
 
 vwn02.pic.es  LINUX       INTEL  Owner      Idle       2.000   263  0+01:55:09
 vwn03.pic.es  LINUX       INTEL  Owner      Idle       2.000   263  0+01:55:10
 
                      Total Owner Claimed Unclaimed Matched Preempting Backfill
 
          INTEL/LINUX     2     2       0         0       0          0        0
 
                Total     2     2       0         0       0          0        0
 
 # chkconfig condor.boot on

Infosystem

gLite CE configuration

We edit /opt/glite/yaim/examples/wn-list.conf We install java

 # apt-get install j2sdk
 Reading Package Lists... Done
 Building Dependency Tree... Done
 The following NEW packages will be installed:
    j2sdk (1.4.2_08-fcs)
 0 upgraded, 1 newly installed, 0 removed and 2 not upgraded.
 Need to get 35.6MB of archives.
 After unpacking 67.5MB of additional disk space will be used.
 Get:1 http://rpm.pic.es slc3/apt/pic j2sdk 2000:1.4.2_08-fcs [35.6MB]
 Fetched 35.6MB in 0s (47.5MB/s)   
 Committing changes...
 Preparing...                ########################################### [100%]
    1:j2sdk                  ########################################### [100%]
 Done
We define the following variables in the site-info.def:
JOB_MANAGER=condor
CE_BATCH_SYS=condor
BATCH_VERSION=condor-6.8.5
We delete line 159 /opt/glite/yaim/scripts/node-info.def
 config_torque_submitter_ssh
We comment line 189 to 192 of /opt/glite/etc/config/scripts/glite-ce-config.py
        #if params['cemon.lrms'] == 'condor' and params['cemon.cetype'] != 'condor':
        #    glib.printErrorMessage("The Condor batch system is supported only with a Condor CE type")
        #    glib.printErrorMessage("Please check the configuration file")
        #    return 1
We change line 3 of /opt/glite/yaim/functions/config_gip_scheduler_plugin from
 requires INSTALL_ROOT CE_BATCH_SYS TORQUE_SERVER
to
 requires INSTALL_ROOT CE_BATCH_SYS
According to the limitation of BALH to use a three letter Batch system name we edit /opt/glite/etc/config/scripts/glite-ce-config.py and add afte (line 960) to
 # Set files mode
 os.chmod("%s/bin/pbs_submit.sh" % os.environ['GLITE_LOCATION'],0755)
 os.chmod("%s/bin/pbs_status.sh" % os.environ['GLITE_LOCATION'],0755)
 os.chmod("%s/bin/pbs_cancel.sh" % os.environ['GLITE_LOCATION'],0755)
 os.chmod("%s/bin/lsf_submit.sh" % os.environ['GLITE_LOCATION'],0755)
 os.chmod("%s/bin/lsf_status.sh" % os.environ['GLITE_LOCATION'],0755)
 os.chmod("%s/bin/lsf_cancel.sh" % os.environ['GLITE_LOCATION'],0755)
the lines
 os.chmod("%s/bin/condor_submit.sh" % os.environ['GLITE_LOCATION'],0755)
 os.chmod("%s/bin/con_status.sh" % os.environ['GLITE_LOCATION'],0755)
 os.chmod("%s/bin/con_cancel.sh" % os.environ['GLITE_LOCATION'],0755)
We modify line 2 in /opt/glite/etc/blah.config from
 supported_lrms=pbs,lsf
to
 supported_lrms=pbs,lsf,condor
and add
####

#Path where CONDOR executables are located
condor_binpath=/opt/condor-local/bin

#Path where the CONDOR conf file is located ($condr_confpath/condor_config)
condor_confpath=/opt/condor-local/etc

#If it is set to no blah scripts for CONDOR will not try to read
#locally from the logs if BLParser is not present
condor_fallback=no

#Set to yes to use Blah Log Parser for CONDOR
condor_BLParser=

#Host where Blah Log Parser for CONDOR is running
condor_BLPserver=127.0.0.1

#Port where Blah Log Parser for CONDOR is running
condor_BLPport=33334

#Number of Blah Log Parser to try for CONDOR (if it is null lsf_BLPserver and lsf_BLPport are used)
condor_num_BLParser=
#
condor_BLPserver1=
condor_BLPport1=
#
condor_BLPserver2=
condor_BLPport2=

We add to /opt/glite/etc/blparser.conf

#
#CONDOR part
#

#Set to yes to start the CONDOR parser
GLITE_CE_USE_BLPARSERCONDOR=yes

#Set to yes to run the CONDOR parser in daemon mode 
GLITE_CE_BLPARSERCONDOR_DAEMON=yes

#Set to 1 for normal debug or 2 for more debug  
GLITE_CE_BLPARSERCONDOR_DEBUGLEVEL=

#Log file where the debug information are written
GLITE_CE_BLPARSERCONDOR_DEBUGLOGFILE=

#CONDOR spooldir
GLITE_CE_BLPARSERCONDOR_SPOOLDIR=

#Use this to start multiple CONDOR parser on different ports
#if it is not set defaults to 1
GLITE_CE_BLPARSERCONDOR_NUM=

#CONDOR parser listening port 
GLITE_CE_BLPARSERCONDOR_PORT1=33334
GLITE_CE_BLPARSERCONDOR_PORT2=

#CONDOR parser listening cream port 
GLITE_CE_BLPARSERCONDOR_CREAMPORT1=
GLITE_CE_BLPARSERCONDOR_CREAMPORT2=
We create the batch system configuration file /opt/glite/etc/batch_gahp.conf
con_config=/opt/condor-local/etc/condor_config
con_binpath=/opt/condor-local/bin

blah scripts

Helper script -- Main.kneuffer - 03 Jul 2007
Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2007-07-05 - unknown
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    EGEE All webs login

This site is powered by the TWiki collaboration platform Powered by Perl This site is powered by the TWiki collaboration platformCopyright & by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Ask a support question or Send feedback