What does 'desired' parameter in autoscaling group do?


#1

With autoscaling on and a parameter already set for min and max nodes, desired seems very amiguous or at least redundant. I cannot seem to find any documentation on this parameter. Can anyone explain?

I’m testing autoscaling on a 4 node cluster (Community edition) and things are not working at all as expected.


#2

Hi @jjv5,

The DesiredCapacity parameter for Auto Scaling Groups is used to inform the AWS platform how many instances the group should maintain. You can refer to the AWS Auto Scaling documentation and the CloudFormation documentation for Auto Scaling Groups for more details. Handling this parameter is not necessary when launching Flight Solo clusters as the template takes care of providing appropriate values for this parameter.

When launching Flight Solo Community Edition via CloudFormation there are two template parameters that dictate how many nodes are made available in your cluster:

  • Initial compute nodes (autoscaling): set this to the number of compute nodes you want in your cluster when it initially starts up.
  • Initial/maximum compute nodes: set this to the maximum number of compute nodes you want your cluster to be able to scale up to.

For a 4-node autoscaling cluster, we’d suggest supplying 1 as the “initial compute nodes” value and 4 as the “maximum compute nodes” value.

HTH, and happy holidays! :christmas_tree: