Skip to content

Latest commit

 

History

History
 
 

example

README.md

MongoDB on Kubernetes with Stateful Sets

With the release of Stateful Sets and custom Storage Classes, Kubernetes can automate all of the underlying infra required to run a MongoDB Replica Sets

Before You Start

  • Have a Kubernetes Cluster created with at least version 1.5.1
  • Have admin access to the cluster

Creating the Storage Class

The storage class will create the Volumes backing the MongoDB Replica Sets.

You can create Storage Classes with different provisioners depending on your Kubernetes environment. There are provisioners for Google Cloud, AWS, Azure, GlusterFS, OpenStack Cinder, vSphere, Ceph RBD, and Quobyte. Pick the right one for your deployment.

For example, this YAML uses a Google Cloud SSD Persistent Disk, and has the name "fast"

Create the Storage Class with the kubectl tool.

kubectl apply -f googlecloud_ssd.yaml

Replace googlecloud_ssd.yaml with another configuration file if you are not using Google Cloud. For example, if you are running Kubernetes on Azure, you can use the Azure SSD YAML. It uses the same name, "fast", so your application does not need to understand the underlying platform.

kubectl apply -f azure_ssd.yaml

If you are using Minikube for local testing:

kubectl apply -f minikube_hostpath.yaml

Verify that the Storage Class is created

$ kubectl get storageclass   
NAME      TYPE
fast       kubernetes.io/gce-pd

Creating the Stateful Set

The example YAML creates a Headless Service and a Stateful Set. It uses the Storage Class created in the previous step, and provisions a 100Gi volume per replica. Modify these values as you see fit.

kubectl apply -f mongo-statefulset.yaml

Verify that the Stateful Set is created

$ kubectl get statefulset
NAME      DESIRED   CURRENT   AGE
mongo     3         0         12m

The Stateful Set controller will spin up each replica one at a time. Eventually, all three will be created.

$ kubectl get statefulset
NAME      DESIRED   CURRENT   AGE
mongo     3         3         12m

You can also verify that the Volumes were created as well.

$ kubectl get pvc         
NAME                               STATUS    VOLUME                                     CAPACITY   ACCESSMODES   AGE
mongo-persistent-storage-mongo-0   Bound     pvc-af87f9d5-d3ab-11e6-8cf2-42010af0018d   100Gi      RWO           12m
mongo-persistent-storage-mongo-1   Bound     pvc-af8cef48-d3ab-11e6-8cf2-42010af0018d   100Gi      RWO           12m
mongo-persistent-storage-mongo-2   Bound     pvc-af8f1d24-d3ab-11e6-8cf2-42010af0018d   100Gi      RWO           12m

Connecting to the MongoDB Replica Set

Each MongoDB Replica Set will have its own DNS address. This will take the format <pod-name>.<service-name>.

For our example, the DNS addresses to use will be:

mongo-0.mongo
mongo-1.mongo
mongo-2.mongo

Put these in your connection url. For example:

mongodb://mongo-0.mongo,mongo-1.mongo,mongo-2.mongo:27017/dbname_?'