In this final functional test, we forcibly removed power from one of the Azure Stack HCI cluster nodes, a7525r06c01n03. This server was running one AKS hybrid worker node VM in the dbaas-databases-1 workload cluster. Approximately four minutes after removing power, the worker node VM was restarted on a7525r06c01n02. All SQL MI pods running on the worker node VM prior to the power outage remained running on the same VM. This scenario is depicted in the following figure.
Figure 37. Azure Stack HCI cluster node failure scenario
Then, we forcibly powered off the worker node VM on a7525r06c01n03. In this case, the SQL MI pods were re-created on another worker node in the dbaas-databases-1 workload cluster. This scenario is depicted in the following figure.
Figure 38. AKS hybrid worker node failure