Operator Not Creating PV In Correct Zone

[brianpham]

• What version of the operator are you running? v0.13.0

• What version of Kubernetes are you running? Running on GKE v1.21.6-gke.1503

Client Version: version.Info{Major:"1", Minor:"22", GitVersion:"v1.22.4", GitCommit:"b695d79d4f967c403a96986f1750a35eb75e75f1", GitTreeState:"clean", BuildDate:"2021-11-17T15:41:42Z", GoVersion:"go1.16.10", Compiler:"gc", Platform:"darwin/amd64"}
Server Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.6-gke.1503", GitCommit:"2c7bbda09a9b7ca78db230e099cf90fe901d3df8", GitTreeState:"clean", BuildDate:"2022-02-18T03:17:45Z", GoVersion:"go1.16.9b7", Compiler:"gc", Platform:"linux/amd64"}

• What are you trying to do? Get the m3db cluster up and running with a persistent volume

• What did you expect to happen? All 3 nodes should come up properly using the correct PV.

• What happened? Only 2 of the 3 nodes are coming up properly. 1 PV is being deployed in us-central1-a, and 2 PV is being deployed in us-central1-b. One PV should be deployed in each region. It seems like operator is ignoring this part of the config

• name: zone-c numInstances: 1 storageClassName: ssd-retain nodeAffinityTerms:

  • key: topology.kubernetes.io/zone values:
    • us-central1-c

This is the config for deploying the cluster

apiVersion: operator.m3db.io/v1alpha1
kind: M3DBCluster
metadata:
  name: m3db-cluster-pv
spec:
  image: quay.io/m3db/m3dbnode:latest
  replicationFactor: 3
  numberOfShards: 1024
  # configMapName: m3db-cluster-pv
  isolationGroups:
  - name: zone-a
    numInstances: 1
    storageClassName: ssd-retain
    nodeAffinityTerms:
    - key: topology.kubernetes.io/zone
      values:
      - us-central1-a
  - name: zone-b
    numInstances: 1
    storageClassName: ssd-retain
    nodeAffinityTerms:
    - key: topology.kubernetes.io/zone
      values:
      - us-central1-b
  - name: zone-c
    numInstances: 1
    storageClassName: ssd-retain
    nodeAffinityTerms:
    - key: topology.kubernetes.io/zone
      values:
      - us-central1-c
  etcdEndpoints:
  - http://etcd-0.etcd:2379
  - http://etcd-1.etcd:2379
  - http://etcd-2.etcd:2379
  namespaces:
  - name: metrics-10s:2d
    preset: 10s:2d
  podIdentityConfig:
    # Using no sources will default to just PodName, which is what we want as
    # remote PVs can move around with the pod
    sources: []
  dataDirVolumeClaimTemplate:
    metadata:
      name: m3db-data
    spec:
      accessModes:
      - ReadWriteOnce
      # this field will be overwritten per-statefulset
      storageClassName: unused
      resources:
        requests:
          storage: 350Gi  

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: ssd-retain
provisioner: kubernetes.io/gce-pd
reclaimPolicy: Retain
parameters:
  type: pd-ssd
allowedTopologies:
- matchLabelExpressions:
  - key: topology.kubernetes.io/zone
    values:
    - us-central1-a
    - us-central1-b
    - us-central1-c

[brianpham]

If I do this, the PV will be created properly. Not sure why the StorageClass above doesn't work though.

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: ssd-retain-us-central1-a
provisioner: kubernetes.io/gce-pd
reclaimPolicy: Retain
parameters:
  type: pd-ssd
allowedTopologies:
- matchLabelExpressions:
  - key: topology.kubernetes.io/zone
    values:
    - us-central1-a
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: ssd-retain-us-central1-b
provisioner: kubernetes.io/gce-pd
reclaimPolicy: Retain
parameters:
  type: pd-ssd
allowedTopologies:
- matchLabelExpressions:
  - key: topology.kubernetes.io/zone
    values:
    - us-central1-b
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: ssd-retain-us-central1-c
provisioner: kubernetes.io/gce-pd
reclaimPolicy: Retain
parameters:
  type: pd-ssd
allowedTopologies:
- matchLabelExpressions:
  - key: topology.kubernetes.io/zone
    values:
    - us-central1-c