nodeSelector does not work(?)

[szymonpk]

I am not sure if I have messed something or it is not fully supported. Config which I am trying

apiVersion: enterprises.upmc.com/v1
kind: ElasticsearchCluster
metadata:
  name: es-cluster
spec:
  kibana:
    enabled: false
    image: docker.elastic.co/kibana/kibana-oss:6.1.3
  cerebro:
    enabled: false
    image: upmcenterprises/cerebro:0.6.8
  elastic-search-image: upmcenterprises/docker-elasticsearch-kubernetes:6.1.3_1
  client-node-replicas: 1
  master-node-replicas: 1
  data-node-replicas: 1
  network-host: 0.0.0.0
  zones: []
  data-volume-size: 20Gi
  java-options: "-Xms512m -Xmx512m"
  snapshot:
    scheduler-enabled: false
    bucket-name: elasticsnapshots99
    cron-schedule: "@every 2m"
    image: upmcenterprises/elasticsearch-cron:0.0.4
  storage:
    storage-class: nfs
  resources:
    requests:
      memory: 512Mi
      cpu: 100m
    limits:
      memory: 1024Mi
      cpu: 100m
  nodeSelector:
    type: muse

ES is installed with helm

helm install --name es0 es-operator/elasticsearch --namespace logging --values elasticsearch/values.yaml --tls

Pods are up, but I got empty result from:

$ kubectl get po es-data-es0-cluster-default-0 -n logging -o yaml | grep nodeSelector

And pods/containers are running on the wrong nodes.

[stevesloka]

This looks like a bug @szymonpk, that field is not implemented anywhere to pass through to the resources created.

[while1eq1]

@szymonpk, Im working on a PR that will implement nodeSelectors / Taints and Tolerations.

[while1eq1]

@szymonpk My PR should offer what your asking for

[while1eq1]

@szymonpk This PR offers nodeSelectors and much more - #270

[szymonpk]

Thanks, much appreciated.

[cfontes]

Just a FYI, there is not much about how this works.

I am trying to use nodeSelector, and I cannot make it work.

My current cluster definition is this

apiVersion: enterprises.upmc.com/v1
kind: ElasticsearchCluster
metadata:
  name: es-cluster
  namespace: observability
spec:
  kibana:
    image: docker.elastic.co/kibana/kibana-oss:6.1.3
  cerebro:
    image: upmcenterprises/cerebro:0.6.8
  nodeSelector:
    cloud.google.com/gke-nodepool: "jaeger-nodepool"
  elastic-search-image: upmcenterprises/docker-elasticsearch-kubernetes:6.1.3_0
  client-node-replicas: 2
  master-node-replicas: 2
  data-node-replicas: 3
  network-host: 0.0.0.0
  data-volume-size: 10Gi
  java-options: "-Xms512m -Xmx512m"
  snapshot:
    scheduler-enabled: false
    bucket-name: elasticsnapshots99
    cron-schedule: "@every 2m"
    image: upmcenterprises/elasticsearch-cron:0.0.4
  storage:
    storage-class-provisioner: kubernetes.io/gce-pd
  zones:
    - europe-west4-c
  resources:
    requests:
      memory: 512Mi
      cpu: 500m
    limits:
      memory: 1024Mi
      cpu: '1'

I never see the nodeselector on the actual pods, it does not work for the tolerations or affinity as well

this is how I am installing the operator

helm repo add es-operator https://raw.githubusercontent.com/upmc-enterprises/elasticsearch-operator/master/charts/
helm install --name elasticsearch-operator es-operator/elasticsearch-operator --set rbac.enabled=True  --namespace observability

then I run this YAML from above using kubectl create -f, I do see everything come up but node selectors are ignore by the operator.

I tried several variations and positions of the node selector in the CRD, but couldn't make it work

[komljen]

@cfontes Make sure that you are using the latest version of the operator. The helm chart is outdated.
You can just pass --set image.tag=0.3.0 to above helm install command.

@stevesloka I think we can close this issue, node tolerations are applied in my test cluster.