Skip to content

Creating a Cluster

Once you've installed the M3DB operator and read over the requirements, you can start creating some M3DB clusters!

Basic Cluster

The following creates an M3DB cluster spread across 3 zones, with each M3DB instance being able to store up to 350gb of data using your Kubernetes cluster's default storage class. For examples of different cluster topologies, such as zonal clusters, see the docs on node affinity.

Etcd

Create an etcd cluster with persistent volumes:

kubectl apply -f https://raw.githubusercontent.com/m3db/m3db-operator/v0.2.0/example/etcd/etcd-pd.yaml

We recommend modifying the storageClassName in the manifest to one that matches your cloud provider's fastest remote storage option, such as pd-ssd on GCP.

M3DB

apiVersion: operator.m3db.io/v1alpha1
kind: M3DBCluster
metadata:
  name: persistent-cluster
spec:
  image: quay.io/m3db/m3dbnode:latest
  replicationFactor: 3
  numberOfShards: 256
  isolationGroups:
  - name: group1
    numInstances: 1
    nodeAffinityTerms:
    - key: failure-domain.beta.kubernetes.io/zone
      values:
      - <zone-a>
  - name: group2
    numInstances: 1
    nodeAffinityTerms:
    - key: failure-domain.beta.kubernetes.io/zone
      values:
      - <zone-b>
  - name: group3
    numInstances: 1
    nodeAffinityTerms:
    - key: failure-domain.beta.kubernetes.io/zone
      values:
      - <zone-c>
  etcdEndpoints:
  - http://etcd-0.etcd:2379
  - http://etcd-1.etcd:2379
  - http://etcd-2.etcd:2379
  podIdentityConfig:
    sources: []
  namespaces:
    - name: metrics-10s:2d
      preset: 10s:2d
  dataDirVolumeClaimTemplate:
    metadata:
      name: m3db-data
    spec:
      accessModes:
      - ReadWriteOnce
      resources:
        requests:
          storage: 350Gi
        limits:
          storage: 350Gi

Ephemeral Cluster

WARNING: This setup is not intended for production-grade clusters, but rather for "kicking the tires" with the operator and M3DB. It is intended to work across almost any Kubernetes environment, and as such has as few dependencies as possible (namely persistent storage). See below for instructions on creating a more durable cluster.

Etcd

Create an etcd cluster in the same namespace your M3DB cluster will be created in. If you don't have persistent storage available, this will create a cluster that will not use persistent storage and will likely become unavailable if any of the pods die:

kubectl apply -f https://raw.githubusercontent.com/m3db/m3db-operator/v0.2.0/example/etcd/etcd-basic.yaml

# Verify etcd health once pods available
kubectl exec etcd-0 -- env ETCDCTL_API=3 etcdctl endpoint health
# 127.0.0.1:2379 is healthy: successfully committed proposal: took = 2.94668ms

If you have remote storage available and would like to jump straight to using it, apply the following manifest for etcd instead:

kubectl apply -f https://raw.githubusercontent.com/m3db/m3db-operator/v0.2.0/example/etcd/etcd-pd.yaml

M3DB

Once etcd is available, you can create an M3DB cluster. An example of a very basic M3DB cluster definition is as follows:

apiVersion: operator.m3db.io/v1alpha1
kind: M3DBCluster
metadata:
  name: simple-cluster
spec:
  image: quay.io/m3db/m3dbnode:latest
  replicationFactor: 3
  numberOfShards: 256
  etcdEndpoints:
  - http://etcd-0.etcd:2379
  - http://etcd-1.etcd:2379
  - http://etcd-2.etcd:2379
  isolationGroups:
  - name: group1
    numInstances: 1
    nodeAffinityTerms:
    - key: failure-domain.beta.kubernetes.io/zone
      values:
      - <zone-a>
  - name: group2
    numInstances: 1
    nodeAffinityTerms:
    - key: failure-domain.beta.kubernetes.io/zone
      values:
      - <zone-b>
  - name: group3
    numInstances: 1
    nodeAffinityTerms:
    - key: failure-domain.beta.kubernetes.io/zone
      values:
      - <zone-c>
  podIdentityConfig:
    sources:
      - PodUID
  namespaces:
    - name: metrics-10s:2d
      preset: 10s:2d

This will create a highly available cluster with RF=3 spread evenly across the three given zones within a region. A pod's UID will be used for its identity. The cluster will have 1 namespace that stores metrics for 2 days at 10s resolution.

Next, apply your manifest:

$ kubectl apply -f example/simple-cluster.yaml
m3dbcluster.operator.m3db.io/simple-cluster created

Shortly after all pods are created you should see the cluster ready!

$ kubectl get po -l operator.m3db.io/app=m3db
NAME                    READY   STATUS    RESTARTS   AGE
simple-cluster-rep0-0   1/1     Running   0          1m
simple-cluster-rep1-0   1/1     Running   0          56s
simple-cluster-rep2-0   1/1     Running   0          37s

We can verify that the cluster has finished streaming data by peers by checking that an instance has bootstrapped:

$ kubectl exec simple-cluster-rep2-0 -- curl -sSf localhost:9002/health
{"ok":true,"status":"up","bootstrapped":true}

Deleting a Cluster

Delete your M3DB cluster with kubectl:

kubectl delete m3dbcluster simple-cluster

After deleting a cluster you'll have to delete the metadata in etcd if you want to reuse the same etcd cluster for a new M3DB cluster. On our roadmap is to add a finalizer to m3dbcluster API objects to allow the operator to clean up their data in etcd. Until then, you'll have to do so manually:

WARNING: This will delete all data in your etcd cluster. If you are using this cluster for anything other than M3DB, you will want to delete only the relevant keys (generally anything containing the string m3db).

kubectl exec etcd-0 -- env ETCDCTL_API=3 etcdctl del --prefix ""