Skip to content

Creating VMs

Create new Azure VMs with azlin's new command - optimized for development workflows with automatic configuration and cloud-init provisioning.

Quick Start

# Create a VM with sensible defaults (128GB RAM)
azlin new

# Create with custom name
azlin new --name my-dev-vm

# Create and clone a repository
azlin new --repo https://github.com/yourusername/yourproject

Overview

The azlin new command provisions Azure VMs with:

  • Pre-installed development tools (Docker, Python, Node.js, Git, etc.)
  • Automatic SSH key management via Azure Key Vault
  • Optional NFS storage mounting for shared home directories
  • Flexible sizing from 8GB to 256GB RAM
  • Template support for repeatable configurations
  • Bastion integration for secure private networking

Command Reference

azlin new [OPTIONS]

Essential Options

Option Description Default
--name TEXT Custom VM name Auto-generated (azlin-vm-xxxxx)
--size [s\|m\|l\|xl] Size tier: s(8GB), m(64GB), l(128GB), xl(256GB) l (128GB)
--vm-size TEXT Exact Azure VM size (overrides --size) Based on size tier
--region TEXT Azure region From config or eastus
--resource-group TEXT Azure resource group From config
--repo TEXT GitHub repository URL to clone None

Advanced Options

Option Description Default
--pool INTEGER Number of VMs to create in parallel 1
--no-auto-connect Skip automatic SSH connection after provision Enabled
--template TEXT Template name for VM configuration None
--nfs-storage TEXT NFS storage account to mount as home None
--no-nfs Skip NFS storage mounting NFS enabled if configured
--no-bastion Always create public IP (skip bastion) Bastion auto-detected
--bastion-name TEXT Explicit bastion host name Auto-detected
-y, --yes Accept all defaults (non-interactive) Interactive

Size Tiers Explained

azlin provides four convenience size tiers that map to optimal Azure VM SKUs:

Tier RAM vCPUs Azure VM Size Use Case
s (small) 8GB 2 Standard_D2s_v3 Lightweight dev, testing
m (medium) 64GB 16 Standard_E16s_v5 Standard development
l (large) 128GB 32 Standard_E32s_v5 Default - ML, large projects
xl (extra-large) 256GB 64 Standard_E64s_v5 Training, heavy workloads

Custom Sizes

Use --vm-size with any Azure VM SKU for exact control: 

azlin new --vm-size Standard_E8as_v5  # ARM-based, cost-effective

Common Usage Patterns

Basic VM Creation

# Default VM (128GB RAM, auto-name, auto-connect)
azlin new

# Small VM for testing (8GB RAM)
azlin new --size s --name test-vm

# Medium VM with custom name (64GB RAM)
azlin new --size m --name dev-environment

What happens:

  1. Generates unique VM name (if not provided)
  2. Creates or retrieves SSH keys from Azure Key Vault
  3. Provisions VM with cloud-init configuration
  4. Installs development tools (Docker, Python, uv, Git, etc.)
  5. Auto-connects via SSH (unless --no-auto-connect)

Development with Repository

# Clone repository during provisioning
azlin new --repo https://github.com/youruser/yourproject

# Large VM for ML project
azlin new --size xl --name ml-training \
  --repo https://github.com/yourteam/ml-pipeline

What happens:

  1. Provisions VM as normal
  2. Clones repository to /home/azureuser/yourproject
  3. Connects you directly to the VM
  4. Repository ready to use immediately

Team Environments with NFS

# Create VM with shared home directory
azlin new --nfs-storage myteam-shared --name worker-1

# Multiple workers sharing same home directory
azlin new --nfs-storage myteam-shared --name worker-2
azlin new --nfs-storage myteam-shared --name worker-3

What happens:

  1. Mounts existing NFS storage account as /home/azureuser
  2. All files, configs, and code shared across VMs
  3. Perfect for distributed workloads or team collaboration

NFS Storage Must Exist

Create NFS storage first with azlin storage create myteam-shared

Parallel VM Provisioning

# Create 5 identical VMs simultaneously
azlin new --pool 5 --size l --name batch-job

# Creates: batch-job-1, batch-job-2, batch-job-3, batch-job-4, batch-job-5

What happens:

  1. Provisions all VMs in parallel (much faster)
  2. Each gets unique name with numeric suffix
  3. All VMs ready simultaneously
  4. No automatic connection (use azlin list to see them)

Template-Based Provisioning

# Create VM from saved template
azlin new --template dev-vm

# Template overrides defaults for size, region, tools, etc.

What happens:

  1. Loads configuration from template
  2. Applies template defaults (size, region, software, etc.)
  3. Any explicit flags override template settings

See: Templates Guide for template creation

Private VMs with Bastion

# Auto-detect bastion (default behavior)
azlin new --name private-vm

# Explicitly specify bastion
azlin new --name private-vm --bastion-name my-bastion

# Force public IP (skip bastion)
azlin new --name public-vm --no-bastion

What happens:

  1. If bastion detected: Creates private VM (no public IP)
  2. If no bastion: Creates VM with public IP
  3. Connection automatically routed through bastion when present

See: Azure Bastion Guide

Non-Interactive (CI/CD)

# Accept all defaults without prompts
azlin new --yes --name ci-runner --size m

# Provision and run command without interaction
azlin new --yes --size xl -- python train.py --epochs 100

What happens:

  1. No interactive prompts
  2. Uses defaults for all unspecified options
  3. Perfect for automation and scripts

Advanced Examples

GPU-Enabled VM

# Specify exact VM size with GPU
azlin new --vm-size Standard_NC6s_v3 --name gpu-training

Multi-Region Deployment

# Deploy to specific regions
azlin new --region eastus --name east-vm
azlin new --region westus --name west-vm
azlin new --region northeurope --name eu-vm

Execute Command Post-Provisioning

# Run command after VM is ready (use -- separator)
azlin new --size xl --repo https://github.com/user/ml-project -- \
  python train.py --config production.yaml

# Install additional software
azlin new --size m -- \
  "sudo apt-get update && sudo apt-get install -y postgresql-client"

Provisioning Time

Typical provisioning times:

  • VM Creation: 3-5 minutes
  • Cloud-init (software): 5-10 minutes
  • Total: ~10-15 minutes for fully configured VM

Watch Progress

Cloud-init logs available via: azlin logs <vm-name>

What Gets Installed

Every azlin VM comes with:

Development Tools: - Python 3.13 (with uv package manager) - Docker & Docker Compose - Node.js & npm - Git & GitHub CLI (gh) - tmux, vim, htop

Cloud Tools: - Azure CLI (authenticated) - kubectl (Kubernetes) - Terraform

System Configuration: - Docker added to user's group (no sudo needed) - SSH keys stored in Key Vault - Automatic security updates enabled

See: Cloud-init Configuration

Troubleshooting

VM Provisioning Fails

# Check Azure quota
azlin quota

# View detailed error logs
azlin logs <vm-name>

# Try different region
azlin new --region westus

Can't Connect After Creation

# Check VM status
azlin status --vm <vm-name>

# Verify SSH keys
az keyvault secret list --vault-name <vault-name>

# Manual connection
azlin connect <vm-name>

Cloud-init Incomplete

# SSH to VM and check cloud-init status
azlin connect <vm-name> -- cloud-init status

# View cloud-init logs
azlin logs <vm-name>

Out of Quota

# Check current quota
azlin quota

# Request quota increase via Azure Portal
# Or try different region/size

azlin new --region westus --size m

See: Quota Management

Source Code

Last updated: 2025-11-24