Microsoft customers were hit by a global outage of the company's Azure virtual machine service -- as well as other Azure services -- that went on about eight hours. Microsoft's Azure status page said the VM outage began around 5 a.m. UTC, or just after midnight ET on October 13.
Around 8 a.m. ET, Microsoft officials said they had mitigated the impact, and that its engineers "proactively verified the return to the full success rate for operations after the updates were completed." Microsoft's status history page says a "subset of customers" using Windows Virtual Machines who attempted to perform service management operations, meaning start, create, update or delete, were affected, but existing running VMs should not have been.
Services with dependencies on Windows VMs, such as Azure DevOps, also may have been affected, the status page noted. (Based on outcry on Twitter, these services were definitely affected.)
Microsoft attributed the issue to "a required artifact version data could not be queried." Microsoft investigated the backend compute resource provider and found a required VMGuestAgent could not be queried from the repository. The VM Guest Agent Extension publishing architecture was in the process of being moved to a new platform using the Azure Resource Manager (ARM) functionality.
Officials said they would work to establish the full root cause in order to head off future occurrences.