Skip to content

cli.md

Latest commit

 

History

History
409 lines (310 loc) · 9.07 KB

cli.md

File metadata and controls

409 lines (310 loc) · 9.07 KB

ModelOps CLI Reference

Overview

The mops command provides infrastructure orchestration for simulation-based methods. Commands are organized into logical groups for managing different aspects of the system.

Global Options

All commands support:

  • --help - Show help for any command

Main Commands

mops infra - Infrastructure Management (Recommended)

Unified infrastructure provisioning that orchestrates all components in correct dependency order.

mops infra init

Generate infrastructure configuration with guided setup.

mops infra init                    # Interactive mode, saves to ~/.modelops/
mops infra init --non-interactive  # Use defaults
mops infra init --output custom.yaml  # Custom location

Options:

  • --output, -o - Custom output path (default: ~/.modelops/infrastructure.yaml)
  • --interactive/--non-interactive - Interactive mode for selections

Behavior:

  • Detects Azure subscriptions from az account list
  • Prompts for subscription selection if multiple found
  • Fetches latest AKS versions for selected region
  • Generates complete infrastructure configuration
  • Saves to ~/.modelops/infrastructure.yaml by default

mops infra up [CONFIG]

Provision infrastructure from YAML configuration.

mops infra up                      # Uses ~/.modelops/infrastructure.yaml
mops infra up infrastructure.yaml   # Custom config
mops infra up --components storage,workspace
mops infra up --env staging --verbose

Arguments:

  • CONFIG - Infrastructure configuration file (optional, defaults to ~/.modelops/infrastructure.yaml)

Options:

  • --components, -c - Specific components to provision (registry,cluster,storage,workspace)
  • --env, -e - Environment name (dev, staging, prod)
  • --verbose, -v - Show detailed output
  • --force, -f - Force reprovisioning even if exists
  • --plan - Preview changes without applying

mops infra down

Destroy infrastructure components.

mops infra down --env dev
mops infra down --env dev --destroy-storage --destroy-registry
mops infra down --env dev --destroy-all --yes

Options:

  • --env, -e - Environment name
  • --destroy-storage - Include storage in destruction
  • --destroy-registry - Include registry in destruction
  • --destroy-all - Destroy all components including data
  • --delete-rg - Also delete resource group (dangerous!)
  • --yes, -y - Skip confirmation prompts
  • --verbose, -v - Show detailed output

mops infra status

Show status of all infrastructure components.

mops infra status
mops infra status --env staging
mops infra status --json

Options:

  • --env, -e - Environment name
  • --json - Output in JSON format

mops infra outputs

Get infrastructure outputs (endpoints, credentials, etc.).

mops infra outputs
mops infra outputs --component cluster
mops infra outputs --show-secrets

Options:

  • --env, -e - Environment name
  • --component - Specific component outputs
  • --show-secrets - Show sensitive values
  • --json - Output in JSON format

mops workspace - Dask Workspace Management

Manage Dask clusters for distributed computation.

mops workspace up

Deploy Dask workspace on existing infrastructure.

mops workspace up --env dev
mops workspace up --config workspace.yaml --env dev

Options:

  • --config - Workspace configuration file
  • --env, -e - Environment name
  • --verbose, -v - Show detailed output

mops workspace down

Destroy Dask workspace.

mops workspace down --env dev
mops workspace down --env dev --force

Options:

  • --env, -e - Environment name
  • --force, -f - Force destroy
  • --verbose, -v - Show detailed output

mops workspace status

Check workspace status and get connection details.

mops workspace status --env dev
mops workspace status --env dev --json

mops jobs - Job Submission

Submit and manage simulation jobs.

mops jobs submit

Submit a simulation job to the cluster.

mops jobs submit examples/study.yaml
mops jobs submit study.yaml --name my-experiment
mops jobs submit study.yaml --output-dir ./results

Arguments:

  • STUDY_FILE - Study configuration file (YAML)

Options:

  • --name - Job name (auto-generated if not provided)
  • --output-dir - Local directory for results
  • --namespace - Kubernetes namespace
  • --env, -e - Environment name
  • --wait - Wait for job completion
  • --follow, -f - Follow job logs

mops jobs list

List all jobs.

mops jobs list
mops jobs list --namespace modelops-dev
mops jobs list --status running

mops jobs status JOB_ID

Get status of a specific job.

mops jobs status job-abc123
mops jobs status job-abc123 --json

mops jobs logs JOB_ID

View job logs.

mops jobs logs job-abc123
mops jobs logs job-abc123 --follow
mops jobs logs job-abc123 --tail 100

mops jobs sync

Sync job status from Kubernetes.

mops jobs sync
mops jobs sync --job-id job-abc123

mops dev - Developer Tools

Developer utilities and testing commands.

mops dev images

Manage Docker image configuration.

# Print image references
mops dev images print scheduler
mops dev images print --all
mops dev images print --profile dev worker

# Export as environment variables
mops dev images export-env
mops dev images export-env --profile local

Actions:

  • print - Print image reference(s)
  • export-env - Export as shell environment variables

Options:

  • --profile - Image profile (prod, dev, local)
  • --all - Show all images

mops dev smoke-test

Run smoke test to verify bundle execution.

mops dev smoke-test
mops dev smoke-test --bundle ./my-bundle
mops dev smoke-test --registry localhost:5000

Options:

  • --bundle - Bundle directory to test
  • --registry - Registry URL
  • --namespace - Kubernetes namespace
  • --verbose, -v - Show detailed output

mops adaptive - Adaptive Optimization

Manage optimization runs (Optuna, MCMC, etc.).

mops adaptive up CONFIG

Start an optimization run.

mops adaptive up optuna-config.yaml
mops adaptive up config.yaml --run-id my-calibration

Arguments:

  • CONFIG - Adaptive configuration file

Options:

  • --run-id - Run identifier (auto-generated if not provided)
  • --env, -e - Environment name
  • --verbose, -v - Show detailed output

mops adaptive down RUN_ID

Stop and clean up an optimization run.

mops adaptive down my-calibration
mops adaptive down my-calibration --force

mops config - Configuration

View and manage ModelOps configuration.

mops config
mops config --json

Shows:

  • Current environment
  • Active profiles
  • Configuration paths
  • Provider settings

mops version - Version Information

Show ModelOps version and component versions.

mops version
mops version --json

Component-Specific Commands (Advanced)

These commands manage individual components. For most use cases, use mops infra instead.

mops cluster - Kubernetes Cluster

  • up - Create AKS cluster
  • down - Destroy cluster
  • status - Check cluster status

mops registry - Container Registry

  • create - Create ACR registry
  • destroy - Destroy registry
  • status - Check registry status
  • login - Authenticate to registry

mops storage - Blob Storage

  • up - Create storage account
  • down - Destroy storage
  • status - Check storage status

Environment Variables

  • MODELOPS_ENV - Default environment (dev, staging, prod)
  • PULUMI_CONFIG_PASSPHRASE_FILE - Pulumi passphrase file location
  • MODELOPS_BUNDLE_REGISTRY - Bundle registry URL
  • AZURE_SUBSCRIPTION_ID - Azure subscription for resources

Configuration Files

Infrastructure Configuration (infrastructure.yaml)

apiVersion: modelops/v1
kind: Infrastructure
metadata:
  name: dev
spec:
  cluster:
    provider: azure
    location: eastus2
    kubernetes_version: "1.30"
    node_count: 2
  storage:
    account_tier: Standard
    account_kind: StorageV2
  registry:
    sku: Basic
  workspace:
    workers:
      replicas: 4
      processes: 2
      threads: 1

Study Configuration (study.yaml)

apiVersion: modelops/v1
kind: Study
metadata:
  name: pi-estimation
spec:
  simulation:
    function: examples.simulations:monte_carlo_pi
    bundle_ref: ""
  parameters:
    n_samples: 100000
  replicates: 100
  seed: 42

Examples

Complete Infrastructure Setup

# 1. Create infrastructure
mops infra up infrastructure.yaml --env dev

# 2. Check status
mops infra status --env dev

# 3. Deploy workspace
mops workspace up --env dev

# 4. Submit a job
mops jobs submit study.yaml

# 5. Check results
mops jobs status <job-id>
mops jobs logs <job-id>

# 6. Clean up
mops workspace down --env dev
mops infra down --env dev --yes

Development Workflow

# Check image configuration
mops dev images print scheduler

# Run smoke test
mops dev smoke-test

# View logs
kubectl logs -n modelops-dask-dev -l app=dask-worker

Exit Codes

  • 0 - Success
  • 1 - General error
  • 2 - Configuration error
  • 3 - Infrastructure error
  • 130 - Interrupted (Ctrl+C)