gitops_pipeline.txt

GITOPS OPERATOR WORK

CI PART
a.	create a preprod dataset in a persistent volume (
	-	unnamespaced: type hostPath
	-	index.html containing text preproduction
	-	StorageClass label "preproduction"
b.	create a prod dataset in a persistent volume 
	-	index.html containing text production
	-	StorageClass label "production"

kubectl apply -f gitops-preprod-pv.yaml
kubectl apply -f gitops-preprod-pvc.yaml
kubectl apply -f gitops-pv-pod-preprod.yaml
kubectl exec -it gitops-pod-preprod -- bash
        echo preproduction > /usr/share/nginx/html/index.html
kubectl delete -f gitops-pod-preprod
kubectl apply -f gitops-prod-pv.yaml
kubectl apply -f gitops-prod-pvc.yaml
kubectl apply -f gitops-pv-pod-prod.yaml
kubectl exec -it gitops-pod-prod -- bash
	echo production > /usr/share/nginx/html/index.html
kubectl delete -f gitops-pod-prod

1.	create a preprod environment as a namespace

	kubectl create ns preprod
	kubectl create ns prod
	# create gitops-preprod git repository and initialize it
	mkdir /gitops-preprod

## verify all of this
##assuming that gitops-preprod has been initialized as a git repo
###
	kubectl create deploy preprod-deploy --image=nginx --replicas=3 --dry-run=client -n preprod -o yaml > preprod-deploy.yaml
	##edit to mount the storage (or use patch)
        ##edit to change label to environment=preproduction
	kubectl apply -f preprod-deploy.yaml
	kubectl expose deploy -n preprod preprod-deploy --label=environment=preproduction --type=NodePort --port=80 --dry-run=client -o yaml > preprod-svc.yaml
	kubectl apply -f preprod-svc.yaml
# at this point we have deploy and svc files in git repo
	kubectl delete -f preprod-svc.yaml
	kubectl delete -f preprod-deploy.svc
	# copy the *pvc.yaml files to current directory
	git add *
	git commit -m "adding initial resources"
	git push
# now we can run gitops-preprod-operator which will do a new kubectl apply every two minutes
	kubectl apply -f gitops-preprod-operator
# now its time to run the script, test and promote to production
# the script should run from a pod in the preprod namespace for smooth access to svc DNS name
	#!/bin/bash
	curl preprod | grep preproduction || exit 3
	cp preprod-deploy.yaml prod-deploy.yaml
	cp preprod-svc.yaml prod-svc.yaml
	sed -i -e 's/environment: preproduction/environment: production/g' prod*.yaml
	## at this point the production files are ready ###VERIFY

	-	poll the git repository for deployments periodically
		-	deployment needs to run an nginx based deployment
		-	comes with a PVC that uses StorageClass as a selector label in the same YAML file
		-	uses label environment=preproduction
	-	expose the application as a nodePort service, using environment=preproduction label
	-	test the application: shell script that greps for preproduction after using curl to connect to the application
	-	shell: if test succeeds
		-	use sed to change environment=preproduction to environment=production in deployment yaml
		-	use sed to change StorageClass name preproduction to production
		-	shell: move this to appropriate Git branch or subdirectory (name=production)



TODO: involve Git Branches

CD PART
2.	create a prod environment as a namespace
	-	poll the git repository for deployments periodically and kubectl apply if there is a diff
	-	expose the application as a nodePort service
	-	test the application
	-	if succeeds, continue running the operator loop