wiki:Cheaha
Last modified 4 years ago Last modified on 10/23/13 13:10:05

Cheaha

Cheaha is an interface to campus high performance computing resources.

Overview

Cheaha Project - The Cheaha cluster information. These pages are mostly historical but have value in recalling early support efforts.
Cheaha Docs - Documented here is Cheaha's history, system profile and scheduling framework

Computation

As an interface to HPC and, in general, the RCS, Cheaha provides an number of entry points. The primary entry point is via SSH. Once on the platform many additional interfaces are available.

DRMAA

The Open Grid Forum developed DRMAA a standard API to distributed resource managers (like SGE and many others). It provides an API that can be embedded into applications that wish to talk to HPC platforms to submit job requests and perform other control functions. Most users will never work with the DRMAA api directly, it is embedded in tools that manage jobs for you. For example, the UAB Galaxy service uses DRMAA on the backend to submit job requests to Cheaha.

While it's not necessary to work with DRMAA directly it can be helpful to test this interface and verify if a front-end application is working correctly. The Son of a Grid Engine project provides a very useful tutorial with example C code.

A working implementation of examples 1 through 2 are available on cheaha via:

git clone ~jpr/git/drmaa-tests.git

Simply make drunjob and then you can submit a sample job for the fictitious command sleeper.sh on the cluster. The command execution will fail but the test will demonstrate successful submission of a job request to SGE via the DRMAA API.

Hardware