Configuring a cluster

Question

Our cluster is a linux cluster with Torque and Maui. I am reading tutorials online and find that the compilation is only on the head node, and after submitting the job by qsub, the other nodes are running the executable compiled on the head node. So that is why my questions arise.

1. Is it correct that a cluster has a head node and many compute nodes. user login and submit jobs at head node via scheduler, the jobs are running on compute nodes not on the head node?

2. can a program be compiled on the head node and run on the other compute nodes? If yes, does it mean that all compute nodes and head node must have same configuration and OS, so a program compiled on the head node can be run on the other nodes without recompiling on each other node?

Answer 1

This depends on what sort of cluster you are talking about, there are many types. In the commercial x86 software market Microsoft have MSCS clusters which is a set of (more or less) equivalently configured Windows server nodes that provide failover capability between each other, NLB Clusters which provide network load balancing and Hyper-V Clusters that support virtualization workloads. VMware have High Availability and Distributed Resource Scheduling Clusters, again to provide various cluster level techniques for clustered Virtual Machines (such as live migration of running VMs between nodes in the cluster). Oracle have their Real Application Clusters (RAC).

And then you have High Performance Computing clusters which allow for the scale out of workloads that can be quantized and run in parallel - there are a bunch of examples - this Wikipedia article isn't a bad place to start exploring from.