Open Cluster Scheduler Go API

Welcome to the Open Cluster Scheduler Go API repository. This project provides a Go-based API to interact with the Open Cluster Scheduler, encapsulating the command-line interface to streamline the configuration and management of Open/Gridware Cluster Scheduler clusters.

Features

The Go API offers a range of powerful features, including:

• qconf Command Line Wrapper: This primary feature enables developers to build robust applications to configure the Open/Gridware Cluster Scheduler effortlessly.
• simulator Application: This application implements a container based simulator of a real "Grid Engine" cluster. It can dump the configuration of a SGE or Open Grid Engine cluster and apply changes to a cluster simulated within a container. (See the cmd/simulator directory for more information.)
• sharetree GUI Editor: This application implements a simple web-based visualization and editing tool for Open Cluster Scheduler (OCS) and Gridware Cluster Scheduler (GCS) sharetree configurations. (See the cmd/sharetree directory for more information.)
• mcp-server: Implements an example MCP (Model Context Protocol) for interacting with the cluster. Process job details, accounting information, cluster configuration details with your favorite AI application (like Claude) which supports MCP extensions.

Go API Development Container

This project includes scripts to set up a one-node Open Cluster Scheduler cluster, which helps you to quickly build and test Go API applications without manually constructing a build environment.

Note, that the tests are written for this container setup. You should not run the tests against a real cluster as they will modify the cluster configuration of course.

Getting Started

Prerequisites

To begin, ensure you have the following software installed:

• Docker
• make tool

Building and Testing Using the Container

Build the Container

First, build the container, which is based on Ubuntu 22.04 and includes all required dependencies:

bash make build

Run the Single Node Cluster

After successfully building the container, you can initialize and run a single-node cluster:

bash make run

Upon the first successful execution of make run, you can expect command line output similar to the following:

shell Install log can be found in: /opt/cs-install/default/common/install_logs/qmaster_install_master_2024-08-11_12:00:00.log Install log can be found in: /opt/cs-install/default/common/install_logs/execd_install_master_2024-08-11_12:00:00.log root@master modified "global" in configuration list root@master modified "all.q" in cluster queue list

This indicates that the cluster has been successfully set up. You can check the status of the cluster by running the following command:

```bash

qhost

HOSTNAME ARCH NCPU NSOC NCOR NTHR LOAD MEMTOT MEMUSE SWAPTO SWAPUS

global - - - - - - - - - - master lx-amd64 7 1 7 7 0.37 23.5G 838.0M 3.0G 0.0 cd /root/go/src/github.com/hpc-gridware/go-clusterscheduler/pkg/qconf ginkgo -v ... ```

The installation directly is the local "installation" subdirectory on the the host which is mounted into the container. The subsequent runs of make run will reuse the existing installation for faster startup. If you want to reinstall the cluster, you can remove the installation directory and run make run again.

Issues

If you encounter any issues or have questions, please open an issue in this repository. We'll be happy to assist.

• No jobs are scheduled: When using a laptop and closing and re-opening, the scheduler thread times out. To protect the qmaster, the scheduler thread gets disconnect. To get it running again either restart the container or remove and attach the scheduler thread manually ("qconf -kt scheduler" and then "qconf -at scheduler").