This project provides kubernetes operators which control backing up and restoring an environment.
It is written in Go using the Kubebuilder framework.
Note: Kubebuilder version 2 is not compatible with the version of OpenShift that Shepherd runs on
Shepherd has integration with this operator. It allows administrators to create new Backup
and Restore
objects via the Drupal UI.
Backup/Restores can be created by creating a new object with a manifest similar to the example below. In this example:
- The
site
andenvironment
labels allow Shepherd to query these objects for display in the UI. volumes
holds a unique name andclaimName
for each PVC that is going to be backed up.mysql
holds a unique name andsecret
for each database that is going to be backed up. Thesecret.keys
property is used by the operator to mount environment variables into the containers running the sql dump/restores in order to connect to the database.
apiVersion: extension.shepherd/v1
kind: Backup
metadata:
name: node-123-backup-xyz
labels:
site: 456
environment: 123
spec:
volumes:
shared:
claimName: node-123-shared
mysql:
default:
secret:
name: node-123
keys:
username: DATABASE_USER
password: DATABASE_PASSWORD
database: DATABASE_NAME
hostname: DATABASE_HOST
port: DATABASE_PORT
status:
startTime: '2018-11-21T00:16:23Z'
completionTime: '2018-11-21T00:16:43Z'
resticId: abcd969xcz
phase: New|InProgress|Failed|Completed
TODO
-
Configure your namespace as required. Used for RBAC rules and defaults to 'myproject' (for local development).
export NAMESPACE=shepherd-dev make kustomize
-
Install the CRD and RBAC.
make install
-
Configure RBAC rules for accounts which should have access to create Backup/Restore objects (i.e. shepherd service account)
oc create clusterrole shepherd-backups --verb=get,list,create,update,delete --resource=backups,restores,syncs oc adm policy add-cluster-role-to-user shepherd-backups --serviceaccount=shepherd
The codebase is written in Go and uses the Kubebuilder framework.
The core logic of this operator is contained in:
- pkg/controller/backup/backup_controller.go in the function
Reconcile()
. - pkg/controller/restore/restore_controller.go in the function
Reconcile()
.
To get started developing this operator, ensure you the following prerequisites:
- A minishift VM running locally
- Go >=1.10 installed
- An IDE such as VSCode or Goland is recommended 👍
- Clone the repo to
$GOPATH/src/github.com/universityofadelaide/shepherd-operator
- Login as cluster admin
oc login -u system:admin
- Run
make install
to set up the CRD in your local OpenShift. - Run
make run
to compile the local workspace and run the operator. Keep it running for the next few steps. - Backup an environment via the Shepherd UI.
- Run
oc get jobs
to check out the jobs that were created.
docker login
to Docker Hub (hub.docker.com).- Set the exnvironment (production in this case) project/namespace
export NAMESPACE=shepeherd-prd
. - Build the image
make docker-build
. - Push the image to Docker Hub
make docker-push
. - Take note of the previous deployment
oc describe sts/shepherd-operator-controller-manager
- Delete the pod and replace it with the new image
oc delete pod shepherd-operator-controller-manager-0
. - Verify
oc logs -f shepherd-operator-controller-manager-0 -c manager
. - Run an upgrade and check that it actually works.