Automated Stale Device Cleanup in Azure Active Directory using a runbook

As with cleaning up inactive guest users, inactive devices also pose several issues for organizations.

Microsoft recommends cleaning up stale devices after 90 days, but does not provide a service option or automation to do so.

Therefore, here’s another runbook you may run to just report on your inactive devices, or to automatically (and optionally periodically) clean up inactive devices in your environment when the removeInactiveDevices switch is supplied.

Managed identity

When run locally, interactive sign in is required. When running as a runbook in Azure automation, the Managed Identity of the automation account is leveraged. This requires you to set Device.ReadWrite.All or Device.Read.All permissions depending on if you want to script to do the cleanup as well.

Autopilot / on premises devices

Note that the script will log an error (and not attempt to delete the device) when a device is an autopilot record (not a real device) or when the device is synced from an on-premises active directory.

Disable vs Delete

The runbook also has a disable option, in which it will first disable a device and wait a configurable ($disableDurationInDays) period of time before actually deleting a device.

Reports

If you wish, you can also let the script mail you a report in CSV format. Add the Mail.Send graph permissions like you did with device permissions and give the MailFrom and MailTo parameters a value.

Download

Download get-AzureADInactiveDevices.ps1 from Gitlab

Disclaimer

As always, the script is provided as-is and should be reviewed and then used at your own risk.

Guest User Last Sign-in date time in Azure Active Directory and automatic cleanup

Azure AD’s sign in logs also only go back 30 days, which makes it highly recommended to stream Azure AD’s sign in logs to a Log Analytics workspace (Azure Monitor). You just need one single P1 license in your tenant to be able to enable this.

However, even if you don’t stream your sign in logs, Microsoft does keep track of when an account last signed in.

My script gets the last sign in data of all guest accounts in your tenant, without any dependencies other than the Az PS module.

If a guest user has never signed in, the creationDate is used to determine inactivity. Otherwise either the last interactive or last non interactive sign in is used (whichever is most recent).

Additionally, the script can also be configured to automatically clean up any guest accounts that have been inactive for a given number of days by using the -removeInactiveGuests switch.

Even in large environments, processing only takes a few minutes at most.

Download

Download the script from my Gitlab here:

https://gitlab.com/Lieben/assortedFunctions/-/blob/master/get-AzureAdInactiveGuestUsers.ps1

Limitations

Microsoft started using these properties in april 2020, so accounts active before that will seem like they have never been active.

Scheduling

This script supports running non-interactive as a runbook in Azure Automation if you supply the -nonInteractive switch. Before this will work, you’ll have to enable Managed Identity on your automation account and run a small script to assign graph permissions to the Managed Identity: AuditLog.Read.All and Organization.Read.All

Reports

If you wish, you can also let the script mail you a report in CSV format. Add the Mail.Send graph permissions like you did with device permissions and give the MailFrom and MailTo parameters a value.

Disclaimer

As always this script is provided as-is and should be reviewed and then used at your own risk.

Onedrive for Business sync error monitoring and auto remediation

With the introduction of Onedrive Sync Heath in the Office portal, we have a much improved view on sync errors of our users. Errors they may not even be aware of.

However, there is no remediation option, so I am sharing a framework based on previous work in Proactive Remediations that can report on the Onedrive client status and trigger a remediation, which looks like this:

Currently, the only remediation method is to restart the Onedrive client, but the script is easily adjusted for additional remediation actions or conditions the community deems useful.

When Paused or Disabled are detected, there is no remediation as this is not technically an error but something the user manually set. This can be adjusted to your local needs easily.

You can find the Proactive Remediation script on Gitlab:

https://gitlab.com/Lieben/assortedFunctions/-/blob/master/remediate-O4BClient.ps1

To use it, follow the steps in my previous article using proactive remediations (LAPS).

the only difference with LAPS config is that the script should run in user context:

The detection and remediation scripts are the same, the script is smart enough to detect if it is running in detection or remediation mode.

Credits go to Rudy Ooms for writing about the different status codes of the Onedrive client 🙂