Possible issues with halyard deployment might require you to help it along with some manual pod and rs management:
Sometimes when deployments fail, new deployments fail due to insufficient memory.
In this case pods get Evicted continuously, and need cleanup:
kubectl get pods -n spinnaker | grep Evicted | awk '{print $1}' | xargs kubectl delete pod -n spinnaker
Find all old (in this example old is v008) pods & rs which haven't scaled down to 0 yet, and scale them down:
kubectl get rs -n spinnaker | grep v008 | awk '{print $1}' | xargs echo
kubectl get rs -n spinnaker | grep v008 | awk '{print $1}' | xargs kubectl scale rs --replicas=0 -n spinnaker