Unable to create a VM using Azure-Client in DevTestLabs. There is a bug in Azure-Cli.

I'm experiencing an issue when attempting to create a VM using a custom image. If I use a standard gallery image, the VM creation command executes successfully. However, when using a custom image, the command consistently fails.

For example, this command works without issue:

az lab vm create --lab-name <LabName> --resource-group <ResourceGroupName> --name <VMName> --image "Ubuntu Server 16.04 LTS" --image-type gallery --size Standard_DS1_v2 --authentication-type ssh --generate-ssh-keys --ip-configuration public

But attempting the following command with a custom image leads to failure:

az lab vm create \

--resource-group NYC \

--lab-name NYC \

--name "NYC-REW" \

--image-type custom \

--image "QWERTY" \

--size Standard_D4s_v3 \

--admin-username azureuser \

--authentication-type ssh \

--generate-ssh-keys \

--ip-configuration public \

--allow-claim false \

--vnet-name "NYC" \

--subnet "JJ_Network_Subnet" \

--os-type "Linux"

This seems to be a bug specific to Azure when handling custom images. Can you please assist with resolving this issue?

Hello Haja Mohideen, instead of relying on the CLI resolving the image by name (which internally tries to find and parse metadata), explicitly provide the full Azure Resource ID of the custom image.

Create a Storage Account and Upload the VHD

STORAGE_NAME=arkvhdstore$RANDOM

LOCATION=eastus

CONTAINER_NAME=vhds

az storage account create \

  --resource-group arkorg \

  --name $STORAGE_NAME \

  --sku Standard_LRS \

  --location $LOCATION

STORAGE_KEY=$(az storage account keys list --account-name $STORAGE_NAME --query '[0].value' -o tsv)

az storage container create \

  --name $CONTAINER_NAME \

  --account-name $STORAGE_NAME \

  --account-key $STORAGE_KEY \

  --public-access off

Download and prepare a base Linux VHD (or use your own)

Create a Temporary VM, Generalize and save as image

az vm create \

  --resource-group arkorg \

  --name tempUbuntuVM \

  --image "Canonical:0001-com-ubuntu-server-jammy:22_04-lts:latest" \

  --admin-username azureuser \

  --authentication-type ssh \

  --generate-ssh-keys \

  --os-disk-name tempUbuntuOSDisk \

  --size Standard_D2s_v3 \

  --location eastus

az vm deallocate --resource-group arkorg --name tempUbuntuVM

az vm generalize --resource-group arkorg --name tempUbuntuVM

az image create \

  --resource-group arkorg \

  --name UbuntuBaseImage \

  --source tempUbuntuVM \

  --os-type Linux

Export VHD from Managed Disk

OSDISK_ID=$(az vm show \

  --resource-group arkorg \

  --name tempUbuntuVM \

  --query "storageProfile.osDisk.managedDisk.id" -o tsv)

SAS_URL=$(az disk grant-access \

  --ids "$OSDISK_ID" \

  --duration-in-seconds 3600 \

  --access-level Read \

  --query accessSas -o tsv)

az storage blob copy start \

  --account-name arkvhdstore3541 \

  --destination-container vhds \

  --destination-blob ubuntu.vhd \

  --source-uri "$SAS_URL" \

  --account-key "$STORAGE_KEY

az storage blob show \

  --account-name arkvhdstore3541 \

  --container-name vhds \

  --name ubuntu.vhd \

  --query "properties.copy.status"

Register Custom Image with DevTest Lab and finally create a DevTest VM from the Custom Image. Here create the VM Using Full Image ID

IMAGE_ID=$(az lab custom-image show \

  --resource-group arkorg \

  --lab-name arkLab \

  --name UbuntuCustomImage \

  --query "id" -o tsv)

az lab vm create \

  --resource-group arkorg \

  --lab-name arkLab \

  --name MyTestVM \

  --image "$IMAGE_ID" \

  --image-type custom \

  --size Standard_D2s_v3 \

  --admin-username azureuser \

  --authentication-type ssh \

  --generate-ssh-keys \

  --ip-configuration public

Once your lab is up, you can push your custom image via the highlighted path. Choose details based on your image type.

Try this and let me know. Hope this resolves your query. Thanks

I don't think you understand the issue. I already have custom images created and can deploy VMs from those images using the Azure portal GUI without any issues. The problem occurs specifically when using Azure CLI—the deployment fails. As I mentioned, this is clearly a bug, and your Azure documentation instructions aren't working as described.