Bulletproof MySQL on Kubernetes using Percona Server
Clone or download

README.md

NOTE: MySQL Operator 0.2.x requires at least Kubernetes 1.11.x (or 1.10.x with alpha features) while version 0.1.x is known to work with Kubernetes up 1.9.x. For upgrading check the 0.2.x upgrade notes since some manual seps are required.

MySQL Operator

MySQL Operator enables bulletproof MySQL on Kubernetes. It manages all the necessary resources for deploying and managing a highly available MySQL cluster. It provides efortless backups, while keeping the cluster highly-available.

MySQL Operator was developed by the awesome engineering team at Presslabs, a Managed WordPress Hosting provider.

For more open-source projects, check Presslabs Code.

Goals and status

The main goals of this operator are:

  1. Easily deploy mysql clusters in kubernetes (cluster-per-service model)
  2. Friendly to devops (monitoring, availability, scalability and backup stories solved)
  3. Out-of-the-box backups (scheduled and on demand) and point-in-time recovery
  4. Support for cloning in cluster and across clusters

The operator is to be considered alpha and not suited for critical production workloads. We (Presslabs) sucessfully use it at the moment for some non-critical production workloads.

Contributing

We welcome all contributions in the form of new issues for feature requests, bugs or directly pull requests. We are open to discuss ideas to improve the operator and would also love to find out where and how it is used. The discussion related to the project should happen on Gitter. The current developers of the project can be reached via email too.

Controller deploy

Install helm. New to helm? Check https://github.com/helm/helm#install

Install kubectl. For more details, see: https://kubernetes.io/docs/tasks/tools/install-kubectl/

To deploy this controller, use the provided helm chart, by running:

helm repo add presslabs https://presslabs.github.io/charts
helm install presslabs/mysql-operator --name mysql-operator

For more information about chart values see chart README. This chart will deploy the controller along with an orchestrator cluster.

v0.2.x upgrade

  1. Scale to 0 the current operator deployment. This won't affect your running databases.
  2. Upgrade your cluster control plane to 1.11
  3. Update the installed CRDs: kubectl apply -f https://raw.githubusercontent.com/presslabs/mysql-operator/master/hack/02x-crds.yaml
  4. Upgrade the mysql-operator helm upgrade mysql-operator presslabs/mysql-operator
  5. Scale the operator deployment up
  6. Now you can upgrade you nodes as well.
  7. Enjoy!

Deploying a cluster

tl;dr

kubectl apply -f https://raw.githubusercontent.com/presslabs/mysql-operator/master/examples/example-cluster-secret.yaml
kubectl apply -f https://raw.githubusercontent.com/presslabs/mysql-operator/master/examples/example-cluster.yaml

More details:

Before creating a cluster, you need a secret that contains the ROOT_PASSWORD key. An example for this secret can be found at examples/example-cluster-secret.yaml.

Create a file named example-cluster-secret.yaml and copy into it the following YAML code:

apiVersion: v1
kind: Secret
metadata:
  name: my-secret
type: Opaque
data:
  # root password is required to be specified
  ROOT_PASSWORD: bm90LXNvLXNlY3VyZQ==

Note.

ROOT_PASSWORD must be base64 encoded.

Now, to create a cluster you need just a simple YAML file that defines it. An example can be found at examples/example-cluster.yaml.

Create a file named example-cluster.yaml and copy into it the following YAML code:

apiVersion: mysql.presslabs.org/v1alpha1
kind: MysqlCluster
metadata:
  name: my-cluster
spec:
  replicas: 2
  secretName: my-secret

To deploy the cluster, run:

kubectl apply -f example-cluster-secret.yaml
kubectl apply -f example-cluster.yaml

For a more in depth configuration, check examples.

To list the deployed clusters use

$ kubectl get mysql
NAME         AGE
my-cluster   1m

To check cluster state use

$ kubectl describe mysql my-cluster
...
Status:
  Ready Nodes:  2
  Conditions:
    Last Transition Time:  2018-03-28T10:20:23Z
    Message:               Cluster is ready.
    Reason:                statefulset ready
    Status:                True
    Type:                  Ready
...

Backups

Backups are stored on object storage services like S3 or google cloud storage. In order to be able to store backup, the secret defined under backupBucketSecretName must the credentials to store those backups. The backups are uploaded using Rclone. The contents of the secret are used to generate an rclone.conf in hack/docker/mysql-helper/docker-entrypoint.sh.

Setup backup to S3

You need to specify the backupBucketUri for the cluster to an uri like s3://BUCKET_NAME, and a secret.

apiVersion: v1
kind: Secret
metadata:
  name: my-cluster-backup-secret
type: Opaque
data:
  AWS_ACCESS_KEY_ID: #
  AWS_SECRET_KEY: #
  # Optional, the AWS region to connect
  # AWS_REGION: us-east1
  # Optional, specify the storage class
  # AWS_STORAGE_CLASS: standard
  # Optional, canned ACL to use
  # AWS_ACL:
  # Optional, the S3 provider to use (default: AWS)
  # S3_PROVIDER: AWS
  # Optional, the S3 endpoint to use (for when you use a different S3_PROVIDER)
  # S3_ENDPOINT:

Then run this command:

apiVersion: v1
kind: Secret
metadata:
  name: my-cluster-backup-secret
type: Opaque
data:
  GCS_SERVICE_ACCOUNT_JSON_KEY: # 
  GCS_PROJECT_ID: #

Then run this command:

kubectl apply -f example-backup-secret.yaml

Note

GCS_SERVICE_ACCOUNT_JSON_KEY and GCS_PROJECT_ID must be base64 encoded.

Requesting a backup

You need to create a file named example-backup.yaml and copy into it the following YAML code:

apiVersion: mysql.presslabs.org/v1alpha1
kind: MysqlBackup
metadata:
  name: my-cluster-backup
spec:
  clusterName: my-cluster

Run the following command:

kubectl apply -f example-backup.yaml 

You need to specify the backupBucketUri for the corresponding cluster to an URI like gs://BUCKET_NAME and backupSecretName. Open the file named example-cluster.yaml and copy into it the following YAML code:

apiVersion: mysql.presslabs.org/v1alpha1
kind: MysqlCluster
metadata:
  name: my-cluster
spec:
  replicas: 2
  secretName: my-secret
  backupSecretName: my-cluster-backup-secret
  backupUri: gs://pl-test-mysql-operator/

Then run the following command:

kubectl apply -f example-cluster.yaml

Listing all backups

$ kubectl get mysqlbackup
NAME                                  AGE
my-cluster-backup                     1m
my-cluster-auto-backup-20180402-1604  1d

Checking the backup state

$ kubectl describe backup my-cluster-backup
...
Status:
  Completed:  true
  Conditions:
    Last Transition Time:  2018-03-21T16:02:56Z
    Message:               
    Reason:                
    Status:                True
    Type:                  Complete
...

Access orchestrator

To connect to orchestrator dashboard you have to port forward orchestrator port 3000 to your local machine. Ensure it's a healthy pod if using raft:

kubectl port-forward mysql-operator-orchestrator-0 3000

Then type localhost:3000 in a browser.

Tech considerations

This project uses Percona Server for MySQL 5.7 because of backup improvements (eg. backup locks), monitoring improvements and some serviceability improvements (eg. utility user). Although we could have used MariaDB, our primary focus being WordPress, we wanted a drop-in rather than a fork. In the future we might support MariaDB if that can be implemented in a compatible way.

License

This project is licensed under Apache 2.0 license. Read the LICENSE file in the top distribution directory, for the full license text.