Schedule Start/Stop for Azure VM with Azure Function

10 Min. Read

Cost management is one of the things to watch out for when managing Azure resources. If you look at the Azure usage report, Azure virtual machine usage would probably be the highest cost contributor to your bill.

As a cost management best practice is to assess whether your virtual machines need to be up and running always or if some can be stopped at a specific schedule to save operational costs. In this case, it may be beneficial to create a schedule to stop and start your Azure virtual machines.

In this article, you will learn how to create an Azure Function where you can automate and schedule the start and stop of your Azure virtual machines using Tags and save costs.

Introduction

Auto-Shutdown Azure VMs by schedule is already a built-in feature in the platform which you can find under the Operations section as shown in the figure below, however, Azure does not have an AutoStart option.

Azure VM Auto-Shutdown
Azure VM Auto-Shutdown

The good news is, Azure provides multiple options to automate and schedule the startup of your virtual machines, and each option has its pros and cons. I will list the different options below so you can decide which option makes sense for you, is scalable, easy to manage, and is cheaper.

Azure Automation

Azure Automation delivers a cloud-based automation and configuration service that supports consistent management across your Azure and non-Azure environments. It comprises process automation, configuration management, update management, shared capabilities, and heterogeneous features. Automation gives you complete control during deployment, operations, and decommissioning of workloads and resources. To automate the schedule of your virtual machines, you create first an automation account and then follow the steps as described in this article.

Automation Tasks (preview)

Microsoft recently announced that you can manage your Azure resources more easily, where you can create automated management tasks for a specific resource or resource group by using automation task templates, which vary in availability based on the resource type. For example, for an Azure storage account, you can set up an automation task that sends you the monthly cost for that storage account. For an Azure virtual machine, you can create an automation task that turns on or turns off that virtual machine on a predefined schedule,  you can follow the steps shown in this video by my dear friend Thomas Maurer.

At the time of this writing, the automation tasks option is in preview, it’s not recommended for production workloads, and is excluded from service level agreements.

Azure Functions (the topic of this article)

With Azure Functions, you can develop more efficiently, an event-driven serverless compute platform that can also solve complex orchestration problems. You can build and debug locally without additional setup, deploy and operate at scale in the cloud, and integrate services using triggers and bindings.

You can read more about Azure Functions here.

How do Automation tasks differ from Azure Automation and Azure Functions?

Automation tasks and Azure Functions are more basic and lightweight than Azure Automation. Currently, you can create an automation task only at the Azure resource level. Behind the scenes, an automation task is actually a logic app resource that runs a workflow and is powered by the multi-tenant Azure Logic Apps service. To use Automation tasks, you need to have an Office 365 account to connect, however, this is not the case with Azure Functions and Azure Automation.

By comparison, Azure Automation is a cloud-based automation and configuration service that supports consistent management across your Azure and non-Azure environments. The service comprises process automation for orchestrating processes by using Runbooks, configuration management with change tracking and inventory, update management for your virtual machines, shared capabilities, and heterogeneous features. Automation account gives you complete control during deployment, operations, and decommissioning of workloads and resources.

In terms of scalability, Azure Functions is more scalable than the Automation task which is configured individually at the Azure resource level that you want to automate, and if you have different schedules that you want to trigger based on, then Azure Automation is not scalable too. With Functions, you write only the code once that truly matters to your business and scale faster.

On the pricing side, I would say that the price for Azure Functions is the cheapest option than using Automation tasks or Azure Automation. Functions app is cheaper when compared to the logic app which is the automation engine behind Automation tasks. When comparing Logic Apps and Azure Functions, both can run under a consumption plan. That means both have a pay-per-usage billing model. I will suggest using the pricing calculator here to compare what would be cheaper between the two options. But if you have high resource utilization then the function will be cheaper than a logic app. Please review the serverless comparison here.

The winner is Azure Functions in terms of scalability, easy to manage for a large number of resources, and cheaper… Read on!

Prerequisites

To follow this article, you need to have the following:

  1. Azure subscription – If you don’t have an Azure subscription, you can create a free one here.
  2. Azure Resource Group (RG) obviously.
  3. Function App with App Service plan consumption-based (more on this in the next section).
  4. Enable system-assigned managed identity for the Function App and assign the right permissions on the desired subscription(s) or management group(s) (more on this in the next section).
  5. At least one Azure virtual machine is deployed in the desired RG. Please check the following quickstart guide to create a Linux or Windows virtual machine.
    • Set the appropriate Tag name and Tag value on the target virtual machine that you want to schedule for auto-shutdown and auto-start (more on this in the next section).

Assuming you have all the prerequisites in place, take now the following steps.

Create Function App

In this section, we’ll create a new function app.

Browse to the Azure Portal and then click Create a resource, select Function App.

Select your desired subscription, resource group, and give it a nice name, such as “stop-start-azvm“.

The runtime stack means what language we would like to write the Function logic. Azure Function provides support for mainstream languages or platforms such as .NET, Java, Node.js, Python, Custom Handler, and PowerShell Core.

For this scenario, we need to choose PowerShell Core as shown in the figure below. We can only choose version 7 which is the latest version at the time of this writing.

Choose your desired Azure region, and then click Next: Hosting > to continue.

Azure Create Function App
Azure Create Function App

Next, we have to create a new or select an existing Storage account, which is required by the latest version of Azure Function V3.

For Plan type, select ‘Consumption‘ for serverless and event-driven scaling for the lowest minimum cost, this Plan is billed dynamically according to your usage, which helps save money when the usage is small as in this case.

As a side note, the ‘Premium‘ is for enterprise-level, serverless applications with event-based scaling and network isolation, or the ‘App Service Plan‘ for reusing compute from the existing app service plan.

Click Next: Monitoring > to continue.

Azure Function Hosting
Azure Function Hosting

On the Monitoring tab, Application Insights can be used to monitor the running health of the function, and can also be used as a console output capture when debugging the function. It’s up to you to decide to enable it or not. In this scenario, I won’t enable it.

Click Review + Create > to review your input, and then select Create.

Azure Function Monitoring
Azure Function Monitoring

After the creation is completed, we need to configure permissions for this Function App to access our virtual machines which could be deployed across multiple subscriptions.

Open your Function App, and then select Identity under the Settings menu, turn on the System assigned toggle to On, then click Save as shown in the figure below, and then confirm by selecting Yes.

Once the system-assigned managed identity is enabled, the Function App resource will be registered with Azure Active Directory (Azure AD). After it’s being registered, we can control its access to other services like storage accounts, subscriptions, or management groups, etc.

Enable system-assigned managed identity
Enable system-assigned managed identity

Now depending on the scope and coverage that you want to give access to the Function App to stop and start your virtual machines. You can give access to a single VM, resource group(s), subscription(s), or management group(s).

In this example, I will give access to my root management group where I have all my subscriptions organized. You need to select your desired scope, then go to Access control (IAM), select Role assignments and then Click + Add | Add role assignment.

Under “Assign access to” choose Function App, search for the Function App name we just created, then add it as a “Virtual Machine Contributor” role as shown in the figure below, and then click Save. The Virtual Machine Contributor role has enough privilege to stop and start the virtual machines.

Add role assignment
Add role assignment

Please note that we need to give that Function App to stop and start VM permission only and not more, please do not give Owner or Contributor role. Always make sure to follow the principle of the least privilege.

Function Code and Trigger

Now we have our Function App up and running, let’s create a new Function.

From the Functions blade under Functions, select + Create to create a Function as shown in the figure below.

Create new function
Create new function

Next, select “Time trigger” as shown in the figure below, and then click Create.

Timer trigger
Timer trigger

After the function is created, go to Integration and then select Time (Timer) under Trigger as shown in the figure below.

Integration Timer
Integration Timer

Next, enter the CRON expression of the timing logic in the “Schedule” field as shown in the figure below. Take 0 0 * * * 1-6 as an example. This means that the function will trigger from Monday to Saturday, every hour (hourly). Click Save.

Set schedule trigger
Set schedule trigger

As a reference, here is the basic format of the CRON expressions in Azure Functions:

{second} {minute} {hour} {day} {month} {day of the week}

e.g. 0 * * * * * (is equal to every minute).

The following values are allowed for the different placeholders:

ValueAllowed ValuesDescription
{second}0-59; *{second} when the trigger will be fired
{minute}0-59; *{minute} when the trigger will be fired
{hour}0-23; *{hour} when the trigger will be fired
{day}1-31; *{day} when the trigger will be fired
{month}1-12; *{month} when the trigger will be fired
{day of the week}0-6; MON-SUN; *{day of the week} when the trigger will be fired

For more information about Azure Functions time trigger, please refer to the official documentation here.

Please keep in mind that the trigger executes at UTC time. So for Bern (UTC+1), a trigger at 18:00 (UTC) executes at 19:00 Switzerland (UTC+1) wintertime and 20:00 (UTC+2) summertime. You will see later in the PowerShell code how we convert the UTC time zone to make sure that we AutoShutdown and AutoStart the VM based on “West Europe Standard Time” and not UTC.

Next, go to Code + Test of the function, and replace the default PowerShell code for run.ps1 with the below code. Please make sure to replace the $subscriptionids variable values with your own Azure subscription(s) information.

# Input bindings are passed in via param block.
param($Timer)

# Add all your Azure Subscription Ids below
$subscriptionids = @"
[
    "aaaaaaaa-1234-5678-0000-999999999999",
    "bbbbbbbb-1234-5678-0000-999999999999",
    "cccccccc-1234-5678-0000-999999999999",
    "dddddddd-1234-5678-0000-999999999999",
    "eeeeeeee-1234-5678-0000-999999999999",
    "ffffffff-1234-5678-0000-999999999999",
    "gggggggg-1234-5678-0000-999999999999",
    "hhhhhhhh-1234-5678-0000-999999999999",
    "iiiiiiii-1234-5678-0000-999999999999",
    "jjjjjjjj-1234-5678-0000-999999999999"
]
"@ | ConvertFrom-Json

# Convert UTC to West Europe Standard Time zone
$date = [System.TimeZoneInfo]::ConvertTimeBySystemTimeZoneId([DateTime]::Now,"W. Europe Standard Time")

foreach ($subscriptionid in $subscriptionids) {
# Selecting Azure Sub
Set-AzContext -SubscriptionId $SubscriptionID | Out-Null

$CurrentSub = (Get-AzContext).Subscription.Id
If ($CurrentSub -ne $SubscriptionID) {
Throw "Could not switch to SubscriptionID: $SubscriptionID"
}

$vms = Get-AzVM -Status | Where-Object {($_.tags.AutoShutdown -ne $null) -and ($_.tags.AutoStart -ne $null)}
$now = $date

foreach ($vm in $vms) {

if (($vm.PowerState -eq 'VM running') -and ( $now -gt $(get-date $($vm.tags.AutoShutdown)) ) ) {
    Stop-AzVM -Name $vm.Name -ResourceGroupName $vm.ResourceGroupName -Confirm:$false -Force
    Write-Warning "Stop VM - $($vm.Name)"
}
elseif (($vm.PowerState -ne 'VM running') -and ( $now -gt $(get-date $($vm.tags.AutoStart)) ) -and ( $now -lt $(get-date $($vm.tags.AutoShutdown)) ) ) {
    Start-AzVM -Name $vm.Name -ResourceGroupName $vm.ResourceGroupName
    Write-Warning "Start VM - $($vm.Name)"
}

}
}

Although param($Timer) in the script above is not used in the code logic, however, it cannot be deleted. This is a required parameter of Timer Trigger.

The Set-AzContext, Get-AzVM, Start-AzVM, and Stop-AzVM cmdlets are not PowerShell’s own modules, they are from the Azure (Az) PowerShell module. Since Azure PowerShell is already installed on the Function host, those commands can be used directly (more on this in a bit).

Click Save to save your code.

Function PowerShell Code
Function PowerShell Code

Now before we test our code, we need to enable the Az PowerShell module on the Function App under the App Service Editor.

From the Functions blade under Development Tools, select App Service Editor (Preview), and then click on Go-> as shown in the figure below.

App Service Editor
App Service Editor

A new browser tab will open where you can edit the requirements for the Function host.

As shown in the figure below, select requirements.psd1 and uncomment the ‘Az’ = ‘6.*’ on line number 7. This is a very important step, otherwise, the function won’t be able to use the Az module, thus the PowerShell code will fail to run.

Enable the use of Az Module in your function app
Enable the use of Az Module in your function app

Close the App Service Editor tab.

Set Tags for Azure VMs

The final step is to set the Tag names and Tag values on the desired virtual machine(s) that you want to schedule for AutoShutdown and AutoStart.

There are multiple ways on how to set and create tags for your Azure resources. In this example, I will show you two options by using the Azure Portal and the Azure PowerShell.

Navigate to your virtual machine in the Azure Portal, on the Overview page, select (change) next to Tags.

As shown in the figure below, enter the Tag names and adjust the value(s) based on your desired schedule. Please make sure the Tag names are entered correctly (AutoShutdown and AutoStart), otherwise, the function won’t work.

In this example, the AutoShutdown is set to 20:00 at night, and AutoStart is set to 07:00 in the morning.

Set tags on Azure VM
Set tags on Azure VM

The good news is, based on the values set on the tags, the Function will automate the stop and start for the VM.

Please note that you can have different values (times) on different virtual machines, and since the function will trigger hourly, it will stop and start the VM based on the values you enter here.

What about if you want to stop and start the VM at half past the hour? In this case, we need to decrease the time trigger and run the function every 30 minutes instead of every hour.

Now if you have many VMs you want to automate this process, we could use the Azure PowerShell to tag our VMs.

Assuming you already have some Tags assigned on your Azure virtual machines, the PowerShell commands below will only add additional tags called AutoShutdown and AutoStart with their respective values “20:00” and “07:00”. For the VMs variable, please replace the “rgname“, and “vmname” with your Azure information.

$vms = @"
[
    {
        "rgname": "resource-group1",
        "vmname": "vm1"
    },
    {
        "rgname": "resource-group2",
        "vmname": "vm2"
    },
    {
        "rgname": "resource-group3",
        "vmname": "vm3"
    }
]
"@ | ConvertFrom-Json

foreach ($vm in $vms) {
    $tags = (Get-AzResource -ResourceGroupName $vm.rgname -Name $vm.vmname).Tags
    $tags += @{ AutoShutdown = "20:00";  AutoStart = "07:00" }
    Set-AzResource -ResourceGroupName $vm.rgname -Name $vm.vmname -ResourceType "Microsoft.Compute/VirtualMachines" -Tag $tags
}

If you want to tag a virtual machine using the Azure CLI, please see How to tag a virtual machine in Azure using the Azure CLI.

Test the Schedule

At this step, we are ready to test our function.

First, you need to change the tag values for AutoShutdown and AutoStart for a VM to a closer hour depends on what time you created the function.

Next, you can trigger the function manually by going to the Function App, select your function, then go to Code + Test, and click “Test/Run” and click Run. If everything is OK you should not see any error in the Logs console as shown in the figure below and now your VM should be stopped or started.

Test and Run
Test and Run

The next option is to simply wait for the next hour where the function will automatically trigger.

If we look at the Activity log for the VM, we can see that schedule worked successfully and the VM was turned off at 20:00 and started at 07:00 in the morning. The event was automatically initiated by the Function App name.

Azure VM Activity log
Azure VM Activity log

That’s there you have it!

Summary

In this article, I showed you how to create an Azure Function so you can schedule the start and stop of your Azure virtual machines based on the defined Tag names and values. This will help you to save costs and you only pay for virtual machines when you’re actually running them.

With Azure Functions, you can build more scalable and stable event-driven applications with a microservices-friendly approach using Functions. Please refer to the pricing pages below for more details:

Logic App Pricing: https://azure.microsoft.com/en-us/pricing/details/logic-apps/

Function App Pricing: https://azure.microsoft.com/en-us/pricing/details/functions/

Azure Automation Pricing: https://azure.microsoft.com/en-us/pricing/details/automation/

As a monthly bonus for the consumption plan pricing, both Azure Functions, and Logic Apps include a monthly free grant* (First 4,000 actions free for logic apps and 400,000 GB-s for functions).

*Free grants apply to paid, consumption subscriptions only. In that scenario, you can leverage the Azure function to automate and schedule the start and stop of your Azure virtual rather than using a logic app (automation tasks) if it is under the free limit.

__
Thank you for reading my blog.

If you have any questions or feedback, please leave a comment.

-Charbel Nemnom-

Related Posts

Previous

Passing the Azure Security Center Ninja Training

Migrating Your Video Pipeline to Azure Media Services

Next

Let me know what you think, or ask a question...

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Subscribe to Stay in Touch

Never miss out on your favorite posts and our latest announcements!

The content of this website is copyrighted from being plagiarized!

You can copy from the 'Code Blocks' in 'Black' by selecting the Code.

Please send your feedback to the author using this form for any 'Code' you like.

Thank you for visiting!