3.1. Using Condor

This section explains the Condor pool configuration on Rocks clusters, and provides link to further documentation.

Machines in the Condor pool can serve a variety of roles:

• Central manager machine is the collector of information, and the negotiator between resources and resource requests. In any Condor pool one machine must be the Central Manager.

• Submit machine allows Condor jobs submission.

• Execute machine provides resources for Condor jobs execution.

The Frontend is configured as the pool's Central Manager, and the Submit machine. The rest of the nodes are configured as Submit/Execute machines.

The configuration of condor is done during the install, the resulting configuration files are located in /opt/condor/etc. To reconfigure condor pool use /opt/condor/sbin/CondorConf command, and then restart condor daemons so the new configuration takes an effect. CondorConf takes -h flag that gives the usage explanation on stdout. Usually, the reconfiguration is only needed when you have multiple network interfaces on your nodes, and the default configuration is not using your desired interface.

By default, on the frontend condor daemons are running on eth1 interface, and on the compute nodes on eth0 interface.

Here is en axample command for configuring condor as the pool manager and the submit machine on the frontend, assuming the forntend has IP 198.202.89.155:

/opt/condor/sbin/CondorConf -n frontend -t sm -m 198.202.89.155

The following command will configure compute nodes as Submit/Execute machines:

/opt/condor/sbin/CondorConf -n compute -t se -m 198.202.89.155

To find information about administrering and using Condor Pools please see the original Condor manual at Condor manuals