Skip to main content

Cluster Walkthrough

This section of the documentation provides a high level view of the organization of various HPCBOX components that make up the cluster.

Majority of the HPCBOX components are installed under /opt/drz/apps/hpcbox/latest. latest is a link to the latest version of HPCBOX which exists under /opt/drz/apps/hpcbox/<version>.

├── etc
├── ext
├── data
├── latest
   ├── clusterapp
   ├── clustercontrol
   ├── hpcbox-common-libs
   ├── utils
   └── hpcboxweb (in preview)

ClusterApp​

HPCBOX ClusterApp is located under /opt/drz/apps/hpcbox/latest/clusterapp.

under the clusterapp directory you will find resources as a subdirectory a few levels below and further below you will find a directory called statics.

cd /opt/drz/apps/hpcbox/latest/clusterapp/latest
cd ./**/**/**/HPCBOX-CLUSTER-APP-linux-x64/resources/app/statics

The statics directory contains all the configuration and integration scripts for HPCBOX ClusterApp.

├── conf
├── docs
├── images
├── js
├── misc
├── scripts
└── tools

conf: ClusterApp configuration files as JSON.

scripts: Application integration scripts.

tools: ClusterApp tool implementations.

js: Helper JavaScript files.

Application Integration Extensions​

BASH script extensions can be added to the out-of-box application integration provided by HPCBOX. These must be placed under /opt/drz/apps/hpcbox/latest/ext/clusterapp/scripts/$DRZ_APP where the value of DRZ_APP must match the tag used by the existing HPCBOX integration for the application. This value is displayed in an Info badge when a application is dragged and dropped into the Workflow Editor.

As an example, in the screenshot below we can see that the DRZAPP value for ANSYS CFD-Post is ANSYSCFDPOST, so any integration extensions must be placed under /opt/drz/apps/hpcbox/ext/clusterapp/scripts/ANSYSCFDPOST

DRZAPP

tip

Make sure your extension script has read and execute permissions for all users on the HPCBOX cluster.

Cluster Control​

The HPCBOX Cluster Control service is part of the Core Platform and is responsible for intelligent autoscaling, alerts and various other platform operations. Cluster Control is installed under /opt/drz/apps/hpcbox/latest/clustercontrol. Log files for the Cluster Control service can be found under /opt/drz/apps/hpcbox/latest/clustercontrol/logs. Log files are dated and there is a new file created for every calendar date. The Cluster Control service can be started/stopped/restarted using the init.d script /etc/init.d/HPCBoxClusterControl.

etc files​

The directory /opt/drz/apps/hpcbox/etc holds the AutoScaler configuration and your HPCBOX Activation file.

Common Libraries​

Libraries which are common to various HPCBOX components are placed under /opt/drz/apps/hpcbox/latest/hpcbox-common-libs/

Utility scripts​

Various utility scripts, like Desktop Manager start scripts, DRM (based on Grid Engine) are placed under /opt/drz/apps/hpcbox/latest/utils

DRM​

Distributed Resource Manager, commonly called queue system on HPC resources are used to handle scheduling of jobs and allotment of resources for jobs on the cluster.

The current DRM used on HPCBOX clusters is based on Grid Engine and is located under /opt/drz/soge. Slurm DRM support is currently under development.