We can add HPC cluster compute nodes to our Condor pools!
- Traditional HPC systems use a batch system such as SGE, PBS or LSF.
- Such batch systems can happily co-exist with Condor.
- Condor executes jobs on all compute nodes:
- SGE/PBS/LSF unaware of Condor;
- whenever SGE runs jobs on a compute node, Condor suspends/vacates.
Requires a little networking sneakiness.