Prepare scale-set workloads for automatic OS disk replacement

Source facts

An automatic OS image upgrade replaces each instance’s operating-system disk, reruns configured extensions and custom-data scripts, and preserves attached data disks. Microsoft requires the image version to be latest. Non-Service Fabric scale sets need application-health monitoring, using either the health extension or a load-balancer probe, not both. For multiple placement groups, the documented probe requirement specifies a Standard Load Balancer. Automatic image upgrades must not be combined with AutomaticByPlatform guest-patch orchestration. Microsoft Learn.

Applicability

Scope this readiness check to Uniform orchestration outside Service Fabric. The documentation identifies automatic OS upgrades for Flexible orchestration as preview and gives that mode additional restrictions. Check the supported image list before choosing the mechanism. Mode and image requirements.

DSE recommendation

Prove that the application can be reconstructed from the scale-set model before enabling automatic image replacement. Ask its owner to identify every change made manually inside an instance, then decide where each required setting or artifact will be maintained. Include bootstrap credentials and the intended data-disk attachment behavior in that review. Choose an application-health response that represents readiness to serve the actual workload.

Verification

Use an approved test scale set to observe an image transition. Compare the resulting application configuration with the declared model, exercise the service transaction, and record the health signal that allowed progression. Resolve missing customizations before expanding the rollout. Preserve upgrade-history evidence alongside the tested image reference; do not treat a deployment flag alone as acceptance.

Official references

Microsoft Learn: Automatic OS image upgrades with Azure Virtual Machine Scale Sets. Source reviewed September 9, 2026.