After Cascade Regional Airlines pushes a new VM image to its booking-site scale set, all instances are replaced simultaneously and the site experiences a brief full outage during the rollout. Which scale set upgrade policy is most likely currently configured, and which change would eliminate the simultaneous-replacement behavior?
Select an answer to reveal the explanation.
Short Explanation
If every instance gets swapped out at the exact same moment, that's Automatic mode doing exactly what it's built to do — apply the change everywhere at once. Rolling mode instead works through the fleet in controlled batches, so there's always a working slice of the site up while the rest updates.
Full Explanation
A scale set's upgrade policy controls how a model change, such as a new image reference, is applied to existing instances: Automatic applies the update to all instances immediately and simultaneously, which is exactly the behavior causing the brief full outage described; Manual requires an administrator to explicitly trigger the upgrade on each instance or batch, giving full control but no automation; and Rolling updates instances in defined batches with health checks and pauses between batches, keeping a portion of the fleet in service throughout the process. Given that the described symptom is a simultaneous full-fleet replacement, the current policy is Automatic, and moving to Rolling is the fix, since Rolling explicitly staggers the update and can be configured with a maximum batch percentage and a pause duration to preserve availability. Switching from Rolling to Automatic, as the first option suggests, would make the outage worse by reintroducing simultaneous replacement, not fix it. Manual mode would not have caused an automatic simultaneous outage in the first place, since it requires explicit administrator action per batch, so it does not match the observed symptom. As an operational check, review the scale set's upgradePolicy.mode property before the next image rollout and confirm it reads Rolling, then verify the rollout's activity log shows batches completing sequentially with health checks passing between them.