High Throughput Computing using Condor

Submitting to the backbone nodes

The Condor backbone nodes remain continuously in the pool unless shut down due to failure or maintenance. The backbone nodes also have more memory and disk storage that the standard nodes as described in the page on understanding the hardware. The backbone nodes are suitable for jobs that can’t run on the standard nodes.

Before submitting jobs to the backbone nodes please read the policy on long running jobs.

Selecting the backbone nodes at submit time requires the HAS_BACKBONE ClassAd, for example by using the following requirements in your submit script:

Requirements = (OpSys == "LINUX" && Arch == "X86_64" && HAS_BACKBONE =?= True)

The

HAS_BACKBONE

ClassAd can be added to, or removed from, unstarted jobs in the queue using our helper scripts add_HASBACKBONE and del_HASBACKBONE as follows.

To add the HAS_BACKBONE ClassAd enter the following at the command line prompt:

add_HASBACKBONE job_identifier

where job_identifier should be replaced by an individual job number, the id for a cluster of jobs, or by your user id if you want to add the ClassAd to all your unstarted jobs.

Similarly to remove the HAS_BACKBONE ClassAd enter:

del_HASBACKBONE job_identifier

where job_identifier should be replaced as described above.

Last modified on May 31, 2017 at 3:32 pm by Pen Richardson