An Azure service that is used to provision Windows and Linux virtual machines.
Azure Windows Server 2016 VM start succeeds but immediately becomes Stopped - SOLUTION FOUND
Solution = All servers have had a ransomware attack.
I am troubleshooting an Azure Windows Server 2016 RDS VM in North Europe that will no longer remain running.
VM details
- OS: Windows Server 2016 Datacenter
- VM size: StandardE4sv3
- Region: North Europe
- Hyper-V generation: V1
- Managed OS disk: Premium_LRS, 256 GB
- VM originally created: December 2017
- Boot Diagnostics: Enabled
Problem
When the VM is started, Azure accepts the operation and the Activity Log records:
Started → Accepted → Succeeded
There is no error or statusMessage associated with the Start operation.
However, the VM does not reach or remain in PowerState/running. After the Start operation completes, Instance View reports:
ProvisioningState/succeeded PowerState/stopped
Starting from PowerState/deallocated produces the same result.
Boot Diagnostics generates a current screenshot showing a black screen.
Troubleshooting already completed
I have tried:
- Multiple controlled Start attempts.
- Full VM deallocation followed by Start.
- Reapply.
- Azure Activity Log records Reapply as:
- Started → Accepted → Succeeded
- VM behaviour is unchanged afterwards.
- Redeploy.
- Cannot be used because the VM cannot remain in the Running state.
- Started → Accepted → Succeeded
- Azure Activity Log records Reapply as:
- Azure Serial Console.
- Azure successfully connects to the VM serial port/COM1.
- No SAC> prompt is presented.
- As the VM was originally created in December 2017, EMS/SAC has not been enabled on the original Windows image.
- Boot Diagnostics.
- Current screenshot reaches a black screen but goes no further.
Compute-VM quota limitation
I attempted to provision a temporary Windows recovery VM so that the snapshot-derived OS disk could be attached and repaired offline.
However, this Azure subscription currently has no available Compute VM-family quota for the required new VM families in North Europe.
Quota increase requests for both ESv5 and Dadsv5 have been rejected by Microsoft, so I currently cannot provision a separate recovery VM.
I also do not want to attach a failed production RDS OS disk to the existing servers.
Question
Given that:
- Azure reports the VM Start operation as successful,
- the VM does not reach the Windows boot logo,
- the VM then remains/returns to PowerState/stopped,
- the Azure VM Agent never becomes Ready,
- Reapply succeeds but makes no difference,
- Redeploy cannot be performed because the VM will not remain running,
- Serial Console connects to COM1 but SAC is unavailable,
- and current Azure Compute quota restrictions prevent creation of a temporary recovery VM,
what is the recommended Microsoft-supported method to recover this existing Windows Server 2016 VM?
Is there a supported method to:
- Enable EMS/SAC without a functioning Azure VM Agent;
- force the VM into Windows Recovery Environment;
- perform an offline repair without provisioning another Azure VM; or
- otherwise diagnose what is causing Windows to power off during the early boot process?
I have retained the original OS disk, an OS-disk snapshot, the snapshot-derived recovery disk, Boot Diagnostics screenshots, Instance View JSON and Azure Activity Log output and can provide these if required.