Imagine a cluster that runs Intel® Cluster Checker on a fixed schedule and even reports failures automatically. By using a few extra tools, some of which may already exist on your system, one could have a self-checking cluster in just a few steps. The key to reach an automatic Intel Cluster Checker is creating a script that will use all these tools.
One of the key benefits of Intel Cluster Checker is that it acts as a real-like user executing commands in the console. The required knowledge to identify what to execute and what to check is then available to anyone.
For instance, to ensure that the coprocessor is up and running, there are three recommended basic steps: two involving the execution of Intel® Many Integrated Core (MIC) Platform Software Stack (MPSS) supporting tools (micinfo and miccheck) plus the execution of a benchmark on the host system that offload work to speed up computation.
This article shows tips and tricks for Intel® Cluster Checker version 2.0.
The tool has switched the main cluster-wide benchmark (used as an application proxy) from HPCC to HPL. This allows a faster but still representative check by default, and can be reconfigured for a stress exercise of the system. The binary is available at /opt/intel/clck/2.0/share/xhpl_intel64_dynamic. It runs an Intel(R) optimized version of the High Performance Linpack* (HPL) benchmark, the actual binary is part of the Intel(R) Math Kernel Library package.
Intel® Cluster Checker version 2.0 provides many new and improved features.
The goal of this version is to include support for third generation Intel® Core™ (codename Ivy Bridge) processors. All the packaged High Performance Computing benchmarks were rebuilt using latest releases of the Intel® Cluster Tools. The list of test modules was streamlined by merging similar checks as well.