temp repo to house the full service configuration while we split it
- Remove kube2iam from kube-system namespace
- change elasticsearch-data-statefulset.yaml to use emptyDir instead of a persistentVolumeClaim:
volumes:
- name: startup
configMap:
name: elasticsearch-startup-1
- name: elasticsearch-data
emptyDir: {}
- Ensure sysctl settings are set on the nodes: vm.max_map_count=262144
You need to run this command a few times to complete the initial deploy. Don't let this get you down, Kubernetes is a convergent system and ordering is too loose to rely on. That's a good thing, it's by design and yields a far more resilient architecture, so feel good each time you run the command and see less errors.
# Run until config has converged
kubectl apply --recursive -f .