Skip to content

Omnia Documentation

Omnia version Downloads Last Commit Contributors Forks License

Omnia is an open-source deployment toolkit for building and managing HPC and AI infrastructure on Linux-based Dell PowerEdge servers. From an Omnia Infrastructure Manager (OIM), administrators can synchronize software content, build stateless node images, discover hardware, provision Slurm and service Kubernetes clusters, deploy telemetry, and run supporting lifecycle utilities.

Modular deployment architecture

Omnia uses a modular, capability-based deployment architecture. Each deployment module owns its configuration, Ansible entry playbook, runtime data, logs, and outputs. Modules exchange documented input/output contracts such as YAML status files, JSON catalogs, and CSV node mappings.

The src/main/omnia.sh script prepares the common runtime and invokes one module at a time. The module's CLI value and source-directory name remain its internal domain identifier. main coordinates setup and execution; it is not a deployment module.

Deployment module Customer outcome Typical dependency
repo_manager Deploy Pulp and synchronize catalog-selected RPMs, images, files, and Python content. None
image_build_manager Deploy image storage services and build functional-group OS images. Successful Repository Manager output
discovery Discover BMC endpoints through OME and produce a PXE mapping. Independent and optional when a mapping is supplied manually
orchestrator Deploy OpenCHAMI and optional OpenLDAP, provision Slurm or service Kubernetes, and optionally start physical nodes through iDRAC PXE boot. Built images, repository information, and a PXE mapping
telemetry Deploy enabled telemetry sources, bridges, and sinks on service Kubernetes. A provisioned service Kubernetes cluster
build_stream Deploy BuildStreaM Manager, GitLab integration, and catalog-driven build and deploy pipelines. Prepared Repository Manager and Image Build Manager services
utils Run utilities such as cluster-log collection, OIM log backup, and unattended OS installation. Depends on the selected utility

For a direct deployment, the source-defined order is:

Repository Manager -> Image Build Manager -> optional Discovery
                   -> Orchestrator -> optional Telemetry

BuildStreaM provides a separate automation path: its build pipeline invokes Repository Manager and Image Build Manager, and its deploy pipeline invokes Orchestrator. Utils is used independently when an operational task requires it.

See Architecture, Running Deployment Modules, and Module Contracts for the detailed interfaces and handoffs.

Choose a deployment path

  • Slurm Quickstart

    Build Slurm images and provision a Slurm cluster.

  • Kubernetes and Telemetry

    Provision service Kubernetes and deploy selected non-LDMS telemetry integrations without Slurm.

  • Full Deployment

    Provision Slurm and service Kubernetes, then deploy the required Telemetry sources and sinks.

  • BuildStreaM

    Use GitLab pipelines to synchronize catalog content, build images, and deploy mapped nodes.

Documentation map

  • Overview

    Learn the architecture, component responsibilities, network topologies, module execution model, and terminology.

  • Get Started

    Select a supported deployment path and follow its required module sequence.

  • How-to Guides

    Complete a task within main, Repository Manager, Image Build Manager, Discovery, Orchestrator, Telemetry, BuildStreaM, or Utils.

  • Reference

    Look up configuration files, module contracts, support matrices, samples, and module playbook entry points.

  • Operations & Maintenance

    Perform day-2 repository, node, BuildStreaM, diagnostic, and platform lifecycle operations.

  • Troubleshooting

    Diagnose cross-module and module-specific failures.

Before deployment

Complete the Prerequisites Checklist, then review the prerequisites for every deployment module included in the selected path. A deployment does not require every module or every optional telemetry source.

Licensing

Omnia is available under the Apache 2.0 license. Omnia deploys open-source and third-party software that remains subject to its own licenses. See the installed software matrix for the applicable components and licenses.

Omnia community members

Dell Technologies Intel University of Pisa Community Member VIZIAS Community Member Liqid

If you have feedback about the Omnia documentation, contact omnia.readme@dell.com.