3 Tips for Working with the Microsoft Azure Recovery Services Agent
In this Ask the Admin, I will show you how to solve some common problems with the Microsoft Azure Recovery Services (MARS) backup agent on servers and Windows devices.
MARS is a cloud backup solution that supports Windows Server and Windows client OSs. It is a good option if you need a simple, no fuss, offsite backup. For more information about MARS and how to install the agent, see Backing Up Files and Folders Using Azure Backup on the Petri IT Knowledgebase.
Connecting to MARS Service
One of the frustrations I have experienced with the MARS agent is when opening the management console snap-in (MMC). MARS times out while connecting to the backend service. The easiest way to avoid this problem is to install the agent while logged into the server or PC with a local administrator account. Using Run as Administrator from a standard user’s desktop is creating the problem. Once the agent is installed and you have registered the device with Azure, you can switch back to a standard, user-account desktop. Finally, you can open the console using Run as Administrator.
If you have not already guessed, you need local admin privileges to access the console. Make sure you follow best practices by removing admin privileges from users. Therefore, users will not have access to Azure Recovery Services on their devices.
The MARS agent does not support local notifications but does write to the Event Log (CloudBackup/Operational) on each device. This can be accessed without any special privileges. Because MARS is primarily intended for backing up server workloads, the lack of local device notifications should not be a surprise. Nevertheless, you will need to check the Recovery Services dashboard in the Azure management portal. The other option is to set up email notifications. These strategies will ensure backup jobs are completing successfully.
On end-user devices, free space can be a particular issue. This is especially true on devices that use low-capacity, solid-state disks (SSDs). It is helpful to understand how the MARS agent backs up files.The first time the MARS agent runs a backup job, it uses a Volume Shadow Copy Service (VSS) snapshot to create a virtual hard disk (VHD) in a scratch folder, and stores checksums for each block of data. You can specify the location of the scratch folder as the agent is installed. On subsequent runs, changed files are isolated using USN Journal tracking and a new VHD is created with an updated file layout. New checksums are compared against the initial checksums to identify changed blocks on files in the current backup set. Any changed blocks are then sent to the Azure vault as blobs. Once the backup job is complete, the differential VHD is merged with the VHD that was created the first time the agent ran on the device, and the checksums in the scratch folder are updated.
To make the initial backup, you will need between 5 percent and 10 percent of free space on the volume. After the initial backup job has completed, the VHD expands and contracts as needed. If you face errors because of a lack of free space on the volume, gradually add folders to the backup set each time the backup runs.
If you are having problems with free space, you can check the amount of space being used by VSS using the vssadmin cmd. This is shown here:
vssadmin list shadowstorage
It is also possible to change the location of the agent scratch folder after installation by modifying the ScratchLocation value in the following registry keys:
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows Azure Backup\Config]
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows Azure Backup\Config\CloudBackupProvider]
In this article, I showed you how to solve some of the common problems with the MARS agent.
More in Backup & Storage
Reviewing Your Backup Checklist
Apr 8, 2022 | Michael Otey
Cloud NAS Shootout: Amazon EFS/FSx versus Azure Files
Apr 4, 2022 | Arian Modiramani
Get Ready for AWS Pi Day 2022
Mar 10, 2022 | Michael Otey
Amazon Adds AWS Backup to S3
Feb 24, 2022 | Michael Otey
Disaster Recovery for Kubernetes
Jan 24, 2022 | Michael Otey
Big Announcements at AWS re:Invent 2021
Dec 2, 2021 | Michael Otey
Most popular on petri