![]() One would think that building an HPC cluster would just be a simple matter of slapping some boxes together with a network and calling it a day. In this final article, I turn my attention back to administration, primarily toward some basic monitoring tools and installation of the final piece of the cluster – the resource manager. ![]() Next, attention was turned to the user by creating and configuring the development and run-time environments and adding Environment Modules, which allow you to control the choice of development tools you use to build and run applications on your cluster. ![]() ![]() In this series on using Warewulf to build an HPC cluster, the first focus was on getting the master node to deploy compute nodes easily and then adding some tools to make the cluster useful (NFS mounts from the master node, user accounts on all of the nodes – recall that the nodes are separate from one another so user accounts aren't on the compute nodes unless you put them there, a parallel shell that allows you to run commands easily on a range of nodes, and NTP on the cluster so the clocks are synchronized).
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |