Service fabric autoscaling RRS feed

  • Question

  • I have added a auto scaling rule in the VM scale set of service fabric. I can see it scale up from default 3 nodes to maximum of 5 nodes when CPU usage is high, and also scale down back to 3 nodes ok when usage drops. However, after it scaled down the nodes, the service fabric cluster still thinks that there should be 5 nodes in the cluster, and shows the two deleted nodes with error status (down). I wonder if this is the expected behavior?
    Tuesday, October 29, 2019 2:27 AM


  • Hi Jerry_Hsi,

    For the service fabric clusters with Bronze durability level, We need to call the Remove-ServiceFabricNodeState cmdlet on our own.

    So for Bronze durability level, This is the expected behavior. For gold and silver durability level, Those node will be removed automatically.

    Its documented in a note here.

    Let me know if your cluster is not in Bronze durability level.

    For the benefit of the community, Please click on "mark as answer" for the replies which helped you to solve the problem

    Tuesday, October 29, 2019 8:25 AM