Deploy Omnia on RHEL Platforms

Follow the below steps to deploy the Omnia container on RHEL-based platforms:

• Space Requirements for the OIM
• Omnia Minimum Cluster Node Requirements
  • Complete Deployment Supported by Omnia (Slurm and Service Kubernetes Cluster) – x86_64 and aarch64
  • Complete Deployment Supported by Omnia (Slurm and Service Kubernetes Cluster) – x86_64 Only
  • Slurm-Only Deployment Supported by Omnia – x86_64
  • Service Kubernetes Cluster-Only Deployment Supported by Omnia (iDRAC Telemetry) – x86_64
• Ports Used by the OIM
  • Container ports
  • OpenCHAMI Ports
  • Telemetry Ports
• Step 1: Deploy Omnia Core Container
  • Prerequisites for Deploying the Omnia Core Container
  • Deploy the Omnia Core Container from Omnia Artifactory
  • Tasks Performed by omnia.sh
  • Access the Omnia Core Container
  • Uninstall Omnia Core Container
  • View Usage Instructions for Omnia Core Container
• Step 2: Create Mapping File with Node Information
  • Create Mapping File
  • Create PXE File Manually
  • Create PXE File Using OME
  • Procedure
  • Groups
  • Functional Groups
  • Recommended Software by Functional Groups
• Step 3: Provide Inputs to the Files in the project_default Directory
  • Input Templates for Deployment
    • Default Values Assumed in these Templates (change if needed):
    • Additional values you must provide:
• Step 4: Provide Required Credentials for Omnia
  • Prerequisites
  • Task Performed by the Playbook
  • Execute the Playbook
  • Things to Keep in Mind
  • Post Execution
• Step 5: Prepare the OIM
  • Prerequisites
  • Procedure
  • network_spec.yml
  • provision_config.yml
  • build_stream_config.yml
  • storage_config.yml
    • Configure PowerScale as S3 Storage
  • Verification
• Step 6: Verify Readiness of OIM
  • Verify PowerScale S3 Connection
  • View Usage Instructions for OpenCHAMI Containers
• Step 7: Configure OpenLDAP Proxy for Centralized Authentication
  • Configure OpenLDAP Connection Type
  • Configure OpenLDAP Proxy Server
• Step 8: Configure Telemetry Requirements
  • Omnia Telemetry Architecture
  • Telemetry Components
  • iDRAC and LDMS Telemetry Data Flows
  • Vector Telemetry Data Flows
  • PowerScale Telemetry Data Flows
  • Vector Telemetry Data Flows
  • PowerScale Telemetry Data Flows
    • Configure Deployment Required for iDRAC Telemetry Service
    • Configure Deployment Required for LDMS Telemetry
    • Configure Deployment Required for PowerScale Telemetry
    • Configure Vector Telemetry Pipeline to Route Data to Victoria from Kafka
    • Telemetry Storage Configuration
• Step 9: Create Local Repositories for the Cluster
  • Prerequisites
  • Input Parameters for Local Repositories
    • /opt/omnia/input/project_default/software_config.json
    • /opt/omnia/input/project_default/local_repo_config.yml
  • Configuring Specific Local Repositories
  • Vast Repo and Vast Client Installation
    • Step 1: Download Vast
    • Step 2: Extract the Package
    • Step 3: Build the Vast Repository
    • Step 4: Host the RPMs on an HTTP Server
    • Step 5: Configure the User Repository in local_repo_config.yml
    • Step 6: Run Omnia Playbooks
  • Execute the Local Repo Playbook
    • Configurations made by the playbook
    • Playbook Execution
    • Metadata Report
    • Check Status of the Packages
    • Log Files
    • Updating Local Repositories after Modifying JSON Files
    • Local Repository Resync
  • Default Admin Debug Packages
  • Add Additional Packages
    • Prerequisites
    • Steps
    • Guidelines for Configuring Additional Packages for Specific Architectures
  • Cleanup Local Pulp Repositories
    • Logs
• Step 10: High Availability
  • Set up High Availability (HA) Kubernetes on the Service Cluster
    • Prerequisites
    • Steps
    • Additional Installations
    • Next Step
• Step 11: Set up Slurm on nodes
  • Automated CUDA and DCGM Provisioning
  • Manual Recovery: CUDA Toolkit and DCGM Setup Failure
    • Add new Slurm nodes
    • Remove Slurm nodes
    • Slurm configuration validation and defaults
  • Configuration merge control
  • Default Slurm configuration
    • Post Installation
  • Pulling container images on a Slurm cluster node
  • HPC Benchmark Image Layer
    • Slurm configuration utilities
  • Backup Slurm configuration
  • Cleanup Slurm configuration
  • Rollback Slurm configuration
• Step 12: Build Cluster Node Images
  • Build Images for x86_64 Cluster Nodes
  • Build Images for aarch64 Cluster Nodes
• Step 13: Provision cluster nodes
  • Prerequisites
  • Update the Input Parameters for Discovery/Provision the Nodes
  • Provision the Cluster Nodes
    • Configurations made by the provision.yml playbook
    • Playbook execution
  • NVIDIA HPC SDK Setup
    • Overview
    • Prerequisites
    • Step 1 — Install on the Compiler Node
    • Step 2 — Set Up on Compute Nodes
    • Environment Variables Configured
    • Verifying the Installation
    • Logs
    • Architecture Support
  • Discover Devices in OpenManage Enterprise and Create Static Groups
    • Prerequisites
    • Procedure
    • BMC Discovery Report
• Step 14: Verify Slurm Cluster and Kubernetes on the Service Cluster
• Step 15: Initialize and Verify Telemetry
  • Prerequisites
  • Steps
  • Collect Telemetry from External Nodes
• Step 16: Verify Telemetry Services Deployed on the Cluster
  • Verify Telemetry Component Status
    • Verify Telemetry-Related Pods Are Running
    • Verify Kubernetes Telemetry Services Attached to Telemetry
  • Verify LDMS Telemetry Flow
    • Verify LDMS Messages in Kafka
    • Verify LDMS Metrics in VictoriaMetrics via Vector
    • Verify Kafka TLS Connectivity
  • Verify iDRAC Telemetry Flow
    • Verify iDRAC Telemetry Messages in Kafka
    • Verify VictoriaMetrics TLS Connectivity
    • View Collected iDRAC Telemetry Data using VictoriaMetrics UI (VMUI) - Cluster Mode Deployment
  • Verify PowerScale Telemetry Flow
    • View Collected PowerScale Telemetry Data using VictoriaMetrics UI (VMUI) - Cluster Mode Deployment
    • View Collected PowerScale Logs using VictoriaLogs UI - Cluster Mode Deployment
  • Accessing the MySQL Database
• External Services Telemetry Integrations
  • Integrate OpenManage Enterprise with Omnia Kafka Pipeline for Secure Telemetry Data Streaming
    • Prerequisites
    • Steps
    • Verify OME Telemetry Data in Kafka
    • Verify OME Telemetry Data in VictoriaMetrics
    • Verify OME Telemetry Data in VictoriaLogs
  • Integrate Smart Fabric Manager (SFM) with VictoriaMetrics for Secure Telemetry Data Streaming
    • Prerequisites
    • Steps
    • View Collected SFM Telemetry Data using VictoriaMetrics UI (VMUI) - Cluster Mode Deployment
  • Integrate NVIDIA Unified Fabric Manager (UFM) with Omnia Telemetry for Secure Metrics and Logs Streaming
    • Supported Metrics and Logs
    • Prerequisites
    • Verify UFM Telemetry Flow
    • View Collected UFM Telemetry Data using VictoriaMetrics UI (VMUI) - Cluster Mode Deployment
    • View UFM Logs using VictoriaLogs
  • Integrate VAST Storage with Omnia Telemetry for Secure Metrics and Logs Streaming
    • Supported Metrics and Logs
    • Prerequisites
    • For VAST Metrics Collection
    • For VAST Log Collection
    • Verify VAST Telemetry Flow
    • View Collected VAST Telemetry Data using VictoriaMetrics UI (VMUI) - Cluster Mode Deployment
    • Key VAST Metrics
    • View VAST Logs using VictoriaLogs
  • Collect Telemetry Data from External Client Nodes to Kafka
    • Prerequisites
    • (Optional) Create a Kafka Topic
    • Extract Kafka connection details and TLS certificates from the Service Kubernetes cluster
    • Produce and Verify Telemetry Data
  • Collect Telemetry Data from External Client Nodes to Victoria DB (Cluster Mode)
    • Prerequisites
    • Retrieve the Victoria Select and Insert LoadBalancer IP Addresses
    • Push Sample metrics from Omnia Core Container in the OIM
  • Collect External Client Nodes to Send Logs to VictoriaLogs (Cluster Mode)
    • Prerequisites
    • Obtain Endpoint Information
    • Configure Syslog Sources
    • Configure HTTP Forwarding Sources
    • Verify Log Ingestion

If you have any feedback about Omnia documentation, please reach out at omnia.readme@dell.com.