Need a dynamic resource allocation option to increase the memory limit for the cluster-autoscaler deployment in PMK.
When the cluster is autoscaled the cluster autoscaler reaches the max limit of memory defined (currently 600MB) for the pods, resulting in OOM kill of the pods and delay in autoscaling cluster.
The workaround is to manually extend the memory limit for the cluster autoscaler deployment on each of the master nodes. However, this change does not persist through a pf9-Kube restart.
Another option is to change the limit in the template file on each master node which will persist through the pf9-Kube service restart but does not persist through host upgrades.