How to Run HA SQL Server on GKE

This post is part of our ongoing series on running Microsoft SQL Server on Kubernetes.  We’ve published a number of articles about running Microsoft SQL Server on Kubernetes for specific platforms and for specific use cases.  If you are looking for a specific Kubernetes platform, check out these related articles.

Running HA SQL Server on Amazon Elastic Container Service for Kubernetes (EKS)

Running HA SQL Server on Azure Kubernetes Service (AKS)

Running HA SQL Server on Red Hat OpenShift

Running HA SQL Server with Rancher Kubernetes Engine (RKE)

Running HA SQL Server on IBM Cloud Private

And now, onto the post…

Google Kubernetes Engine (GKE) is a managed, production-ready environment for deploying containerized applications in Google Cloud Platform. Launched in 2015, GKE is one of the first hosted container platforms which is built on the learnings from Google’s experience of running services like Gmail and YouTube in containers for over 12 years. GKE allows customers to quickly get up and running with Kubernetes by completely eliminating the need to install, manage, and operate Kubernetes clusters.

Portworx is a cloud native storage and data management platform to run persistent workloads deployed on a variety of orchestration engines including Kubernetes. With Portworx, customers can manage the database of their choice on any infrastructure using any container scheduler. It provides a single data management layer for all stateful services, no matter where they run.

This tutorial is a walk-through of the steps involved in deploying and managing a highly available Microsoft SQL Server database on Google Kubernetes Engine. Portworx is a Microsoft SQL Server high availability and disaster recovery partner and this tutorial will show you how to reliably run this database for mission-critical applications.

In summary, to run HA SQL Server on Google Cloud Platform you need to:

1. Launch a GKE cluster
2. Install cloud native storage solution like Portworx as a DaemonSet on GKE
3. Create a storage class defining your storage requirements like replication factor, snapshot policy, and performance profile
4. Deploy SQL Server on Kubernetes
5. Test failover by killing or cordoning node in your cluster
6. Expand the storage volume without downtime
7. Backup and restore SQL Server from a snapshot

While not covered in this tutorial, Portworx also enables the following capabilities for MS SQL server.

• Bring-your-own key encryption of data volumes for data protection
• Zero RPO failover of MS SQL server pods between data centers in a metropolitan area (defined by maximum 15 milliseconds round trip latency)
• Near RPO zero failover for MS SQL server pods between data centers across the WAN
• Application and data backup and restore across environments with a single Kubernetes command for scheduled and unscheduled maintenance, blue-green deployments or copy data management.

How to set up a GKE cluster

When launching a GKE cluster to run Portworx, you need to ensure that the cluster is based on Ubuntu. Due to certain restrictions with GKE clusters based on Container-Optimized OS (COS), Portworx requires Ubuntu as the base image for the GKE Nodes.

The following command configures a 3-node GKE Cluster in zone ap-south-1-a. You can modify the parameters accordingly.

$ gcloud container clusters create "gke-px" \
	--zone "asia-south1-a" \
	--username "admin" \
	--cluster-version "1.11.8-gke.6" \
	--machine-type "n1-standard-4" \
	--image-type "UBUNTU" \
	--disk-type "pd-ssd" \
	--disk-size "50" \
	--scopes "https://www.googleapis.com/auth/compute","https://www.googleapis.com/auth/devstorage.read_only","https://www.googleapis.com/auth/logging.write","https://www.googleapis.com/auth/monitoring","https://www.googleapis.com/auth/servicecontrol","https://www.googleapis.com/auth/service.management.readonly","https://www.googleapis.com/auth/trace.append" \
	--num-nodes "3" \
	--enable-cloud-logging \
	--enable-cloud-monitoring \
	--network "default" \
	--addons HorizontalPodAutoscaling,HttpLoadBalancing,KubernetesDashboard

Once the cluster is ready, configure kubectl CLI with the following command:

$ gcloud container clusters get-credentials gke-px --zone asia-south1-a

Portworx requires a ClusterRoleBinding for your user. Without this configuration, the command fails with an error clusterroles.rbac.authorization.k8s.io "portworx-pvc-controller-role" is forbidden.

Let’s create a ClusterRoleBinding with the following command:

$ kubectl create clusterrolebinding cluster-admin-binding \
--clusterrole cluster-admin \
--user $(gcloud config get-value account)

You should now have a three node Kubernetes cluster deployed in Google Cloud Platform.

$ kubectl get nodes
NAME                                    STATUS   ROLES    AGE   VERSION
gke-gke-px-default-pool-2d6ef8b8-41w3   Ready    <none>   1d    v1.11.8-gke.6
gke-gke-px-default-pool-2d6ef8b8-gd1d   Ready    <none>   1d    v1.11.8-gke.6
gke-gke-px-default-pool-2d6ef8b8-pn9n   Ready    <none>   1d    v1.11.8-gke.6

Installing Portworx in GKE

Installing Portworx on GKE is not very different from installing it on any other Kubernetes cluster. Portworx GKE documentation has the steps involved in running the Portworx cluster in a Kubernetes environment deployed in GCP.

Once the GKE cluster is up and running, and Portworx is installed and configured, we will deploy a highly available SQL Server instance.

The Portworx cluster needs to be up and running on GKE before proceeding to the next step. The kube-system namespace should have the Portworx pods in Running state.

$ kubectl get pods -n=kube-system -l name=portworx
NAME             READY   STATUS    RESTARTS   AGE
portworx-7q6dh   1/1     Running   1          1d
portworx-d9v6t   1/1     Running   1          1d
portworx-lfrjn   1/1     Running   1          1d

Creating a storage class for MS SQL Server

Once the GKE cluster is up and running, and Portworx is installed and configured, we will deploy a highly available Microsoft SQL Server stack in Kubernetes.

Through storage class objects, an admin can define different classes of Portworx volumes that are offered in a cluster. These classes will be used during the dynamic provisioning of volumes. The storage class defines the replication factor, I/O profile (e.g., for a database or a CMS), and priority (e.g., SSD or HDD). These parameters impact the availability and throughput of workloads and can be specified for each volume. This is important because a production database will have different requirements than a development Jenkins cluster.

$ cat > px-sql-sc.yaml << EOF
kind: StorageClass
apiVersion: storage.k8s.io/v1beta1
metadata:
    name: px-mssql-sc
provisioner: kubernetes.io/portworx-volume
parameters:
   repl: "3"
   io_profile: "db_remote"
   priority_io: "high"
allowVolumeExpansion: true 
EOF

Create the storage class and verify it’s available in the default namespace.

$ kubectl create -f px-sql-sc.yaml
storageclass.storage.k8s.io/px-mssql-sc created

$ kubectl get sc
NAME                 PROVISIONER                     AGE
px-mssql-sc          kubernetes.io/portworx-volume   11s
standard (default)   kubernetes.io/gce-pd            1d
stork-snapshot-sc    stork-snapshot                  1d

Creating a MS SQL Server PVC

We can now create a Persistent Volume Claim (PVC) based on the Storage Class. Thanks to dynamic provisioning, the claims will be created without explicitly provisioning Persistent Volume (PV).

$ cat > px-sql-pvc.yaml << EOF
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
   name: mssql-data
   annotations:
     volume.beta.kubernetes.io/storage-class: px-mssql-sc
spec:
   accessModes:
     - ReadWriteOnce
   resources:
     requests:
       storage: 5Gi
EOF

$ kubectl create -f px-sql-pvc.yaml
persistentvolumeclaim/mssql-data created

Let’s verify the PVC with the following command:

$ kubectl get pvc
NAME         STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS   AGE
mssql-data   Bound    pvc-8559f499-63fc-11e9-b0dd-42010aa0004a   5Gi        RWO            px-mssql-sc    40s

Deploying MS SQL Server on GKE

Finally, let’s create a Microsoft SQL Server instance as a Kubernetes deployment object. For simplicity sake, we will just be deploying a single SQL Server pod. Because Portworx provides synchronous replication for High Availability, a single SQL Server instance might be the best deployment option for your SQL database. Portworx can also provide backing volumes for multi-node SQL Server deployments. The choice is yours.

cat > px-sql-db.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
  name: mssql
spec:
  strategy:
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 1
    type: RollingUpdate
  replicas: 1
  selector:
    matchLabels:
      app: mssql
  template:
    metadata:
      labels:
        app: mssql
    spec:
      containers:
      - name: mssql
        image: microsoft/mssql-server-linux:2017-latest
        imagePullPolicy: "IfNotPresent"
        ports:
        - containerPort: 1433
        env:
        - name: ACCEPT_EULA
          value: "Y"
        - name: SA_PASSWORD
          value:  "P@ssw0rd"
        volumeMounts:
        - mountPath: /var/opt/mssql
          name: mssqldb
      volumes:
      - name: mssqldb
        persistentVolumeClaim:
          claimName: mssql-data
EOF

$ kubectl create -f px-sql-db.yaml
deployment.apps/mssql created

Make sure that the SQL Server pods are in the Running state.

$ kubectl get pods -l app=mssql -o wide --watch

Wait till the SQL Server pod reaches the Running state.

$ kubectl get pods -l app=mssql -o wide --watch
NAME                     READY   STATUS    RESTARTS   AGE   IP          NODE                                    NOMINATED NODE
mssql-79759f8f66-7zzx2   1/1     Running   0          63s   10.52.2.9   gke-gke-px-default-pool-2d6ef8b8-41w3   

We can inspect the Portworx volume by accessing the pxctl tool running with the SQL Pod.

$ VOL=`kubectl get pvc | grep mssql-data | awk '{print $3}'`
$ PX_POD=$(kubectl get pods -l name=portworx -n kube-system -o jsonpath='{.items[0].metadata.name}')
$ kubectl exec -it $PX_POD -n kube-system -- /opt/pwx/bin/pxctl volume inspect ${VOL}

The output from the above command confirms the creation of volumes that are backing the SQL Server database instance.

Failing over MS SQL Server on GKE

Let’s populate the database with sample data.

Create a SQL file with the below statements:

CREATE DATABASE classicmodels;
go

USE classicmodels;
go

CREATE TABLE offices (
  officeCode varchar(10) NOT NULL,
  city varchar(50) NOT NULL,
  phone varchar(50) NOT NULL,
  addressLine1 varchar(50) NOT NULL,
  addressLine2 varchar(50) DEFAULT NULL,
  state varchar(50) DEFAULT NULL,
  country varchar(50) NOT NULL,
  postalCode varchar(15) NOT NULL,
  territory varchar(10) NOT NULL,
);
go

insert  into offices(officeCode,city,phone,addressLine1,addressLine2,state,country,postalCode,territory) values
('1','San Francisco','+1 650 219 4782','100 Market Street','Suite 300','CA','USA','94080','NA'),
('2','Boston','+1 215 837 0825','1550 Court Place','Suite 102','MA','USA','02107','NA'),
('3','NYC','+1 212 555 3000','523 East 53rd Street','apt. 5A','NY','USA','10022','NA'),
('4','Paris','+33 14 723 4404','43 Rue Jouffroy abbans',NULL,NULL,'France','75017','EMEA'),
('5','Tokyo','+81 33 224 5000','4-1 Kioicho',NULL,'Chiyoda-Ku','Japan','102-8578','Japan'),
('6','Sydney','+61 2 9264 2451','5-11 Wentworth Avenue','Floor #2',NULL,'Australia','NSW 2010','APAC'),
('7','London','+44 20 7877 2041','25 Old Broad Street','Level 7',NULL,'UK','EC2N 1HN','EMEA');
go

We will copy the sample data to the MS SQL pod before loading it through the SQLCMD utility.

$ SQL_POD=$(kubectl get pods -l app=mssql -o jsonpath='{.items[0].metadata.name}')
$ kubectl cp sample_data.sql $SQL_POD:/tmp

Let’s load the sample data into SQL Server.

$ kubectl exec $SQL_POD -- /opt/mssql-tools/bin/sqlcmd -U sa -P P@ssw0rd -i /tmp/sample_data.sql

We can query the database by running the below command:

$ kubectl exec $SQL_POD -- /opt/mssql-tools/bin/sqlcmd -U sa -P P@ssw0rd -d classicmodels -Q 'select * from offices'

The below query shows only the cities from the table.

$ kubectl exec $SQL_POD \
-- /opt/mssql-tools/bin/sqlcmd -U sa -P P@ssw0rd \
-d classicmodels -Q 'select city from offices'

Now, let’s simulate the node failure by cordoning off the node on which SQL Server is running.

$ NODE=`kubectl get pods -l app=mssql -o wide | grep -v NAME | awk '{print $7}'`
$ kubectl cordon ${NODE}
node/gke-gke-px-default-pool-2d6ef8b8-41w3 cordoned

We will now go ahead and delete the SQL Server pod.

$ POD=`kubectl get pods -l app=mssql -o wide | grep -v NAME | awk '{print $1}'`
$ kubectl delete pod ${POD}
pod "mssql-79759f8f66-7zzx2" deleted

As soon as the pod is deleted, it is relocated to the node with the replicated data. STorage ORchestrator for Kubernetes (STORK), Portworx’s custom storage scheduler allows co-locating the pod on the exact node where the data is stored. It ensures that an appropriate node is selected for scheduling the pod.

Let’s verify this by running the below command. We will notice that a new pod has been created and scheduled in a different node.

$ kubectl get pods -l app=mssql
NAME                     READY   STATUS              RESTARTS   AGE
mssql-79759f8f66-l7zjf   0/1     ContainerCreating   0          37s

Wait till the pod is ready and run the select query in it.

$ SQL_POD=$(kubectl get pods -l app=mssql -o jsonpath='{.items[0].metadata.name}')
$ kubectl exec $SQL_POD \
-- /opt/mssql-tools/bin/sqlcmd -U sa -P P@ssw0rd \
-d classicmodels -Q 'select city from offices'

Observe that the database table is still there and all the content intact!

Performing Storage Operations on Kubernetes

After testing end-to-end failover of the database, let’s perform StorageOps on our GKE cluster.

Expanding the Volume with no downtime

Let’s first get the Portworx volume name backing the SQL Server deployment and inspect it through the pxctl tool.

$ VOL=`kubectl get pvc | grep mssql-data | awk '{print $3}'`
$ PX_POD=$(kubectl get pods -l name=portworx -n kube-system -o jsonpath='{.items[0].metadata.name}')
$ kubectl exec -it $PX_POD -n kube-system -- /opt/pwx/bin/pxctl volume inspect ${VOL}

The current volume size is 5GiB as defined in the PVC specification. Let’s expand it to 7GiB using the following command.

$ kubectl exec -it $PX_POD -n kube-system --  /opt/pwx/bin/pxctl volume update $VOL --size=7
Update Volume: Volume update successful for volume pvc-8559f499-63fc-11e9-b0dd-42010aa0004a

Backing up and restoring a SQL Server instance through snapshots

Portworx supports creating Snapshots for Kubernetes PVCs. Since there is only one SQL Server instance, we can use regular, local snapshots to backup and restore.

Let’s create a snapshot for the Kubernetes PVC we created for SQL Server.

cat >  px-sql-snap.yaml << EOF
apiVersion: volumesnapshot.external-storage.k8s.io/v1
kind: VolumeSnapshot
metadata:
  name: px-sql-snapshot
  namespace: default
spec:
  persistentVolumeClaimName: mssql-data
EOF

$ kubectl create -f px-sql-snap.yaml
volumesnapshot.volumesnapshot.external-storage.k8s.io/px-sql-snapshot created

Verify the creation of the volume snapshot.

$ kubectl get volumesnapshot
NAME                AGE
px-sql-snapshot   25s

$ kubectl get volumesnapshotdatas
NAME                                                       AGE
k8s-volume-snapshot-251628c0-6406-11e9-85ba-0a580a340105   33s

We can now create a new PVC from the snapshot.

$ cat > px-sql-snap-pvc.yaml << EOF
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: px-sql-snap-clone
  annotations:
    snapshot.alpha.kubernetes.io/snapshot: px-sql-snapshot
spec:
  accessModes:
     - ReadWriteOnce
  storageClassName: stork-snapshot-sc
  resources:
    requests:
      storage: 5Gi
EOF

$ kubectl create -f px-sql-snap-pvc.yaml
persistentvolumeclaim/px-sql-snap-clone created

A new SQL Server pod based on the new PVC restored from the snapshot will contain the data from the original volume.

$ cat > px-sql-db-clone.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
  name: mssql-snap
spec:
  strategy:
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 1
    type: RollingUpdate
  replicas: 1
  selector:
    matchLabels:
      app: mssql-snap  
  template:
    metadata:
      labels:
        app: mssql-snap
    spec:
      containers:
      - name: mssql
        image: microsoft/mssql-server-linux:2017-latest
        imagePullPolicy: "IfNotPresent"
        ports:
        - containerPort: 1433
        env:
        - name: ACCEPT_EULA
          value: "Y"
        - name: SA_PASSWORD
          value:  "P@ssw0rd"
        volumeMounts:
        - mountPath: /var/opt/mssql
          name: mssqldb
      volumes:
      - name: mssqldb
        persistentVolumeClaim:
          claimName: px-sql-snap-clone
EOF

 
$ kubectl create -f px-sql-db-clone.yaml
deployment.apps/mssql-snap created

Querying the new SQL Server pod will show the same data as the original.

$ SQL_POD=$(kubectl get pods -l app=mssql-snap -o jsonpath='{.items[1].metadata.name}')
$ kubectl exec $SQL_POD \
> -- /opt/mssql-tools/bin/sqlcmd -U sa -P P@ssw0rd \
> -d classicmodels -Q 'select city from offices'
city
--------------------------------------------------
San Francisco
Boston
NYC
Paris
Tokyo
Sydney
London

(7 rows affected)

Summary

Portworx can easily be deployed on Google GKE to run stateful workloads in production. It integrates well with K8s StatefulSets by providing dynamic provisioning. Additional operations such as expanding the volumes and performing backups stored as snapshots on object storage can be performed while managing production workloads.