Helm is a popular package manager for Kubernetes that is leveraged to deploy the UDMIF application to the Kubernetes cluster that is created as part of the terraform process. The UDMIF API and Web source code needs to be built locally and then deployed using the appropriate helm charts and scripts.
There are some prerequisites that need to be satisfied in order to build the docker images for those projects and deploy them to Kubernetes.
The API and Web folders each contain a docker build script that generates the docker image for the application and publish it to GCR.
NOTE: Make sure your gcloud CLI is authenticated against the project you will be using in GCP and properly configured.
gcloud config set project <project_id> gcloud config set compute/zone <region> gcloud auth login gcloud auth configure-docker
Here is an example on how to run the docker build script:
# ./buildDocker.sh -p <projectId> -t <image-tag>
./buildDocker.sh -p my-project-id -t latest
Helm is used to template and deploy the Web and API UDMIF services in GKE. You can get more information about Helm right here.
You need to configure your project ID, compute zone and GKE cluster credentials before you can deploy pods in it.
Example:
gcloud container clusters get-credentials <cluster_name>
Deploy the helm charts using helm upgrade commands for the appropriate image tag you built and deployed above. You need to update the GCP_PROJECT_ID in the repository: section and provide values for all variables in the env: section for API and Web helm charts before running the commands. Once the values.yaml file has been updated, you can install or update the chart with the following commands.
You need to be in the helm directory before issuing the commands. Image Tag should be a tag that was pushed to GCR in the docker build phase.
udmi-web:
helm upgrade --install --debug --namespace udmi --create-namespace --set image.tag=<image tag> udmi-web ./udmi-web
udmi-api:
helm upgrade --install --debug --namespace udmi --create-namespace --set image.tag=<image tag> udmi-api ./udmi-api
Deploy the ingress using the following command. Please note that you must update the HOST_NAME in ingress.yaml before running the command. that hostname should be using the full domain name that you created in terraform. For example: dashboard.udmi.mydomain.com. You should ensure that you have proper DNS records in place so that this domain is resolvable over the internet.
kubectl apply -f ingress.yaml --namespace udmi
eventHandler cloud function.The event handler cloud function is deployed with a script that leverages gcloud to create the function. The script is named [deploy.sh] (event-handler/deploy.sh) located in the event-handler folder. Please ensure you set the required environment variables before running the script. You also need to ensure your credential.json file is available at the proper location that your project ID is set.
example:
#run from the event-handler folder.
export REGION=us-central1
export POSTGRESQL_DATABASE=udmi
export POSTGRESQL_PORT =5432
export POSTGRESQL_USER=someUser
export POSTGRESQL_PASSWORD=somePassword
export POSTGRESQL_INSTANCE_HOST=privateIP
#Authentication with GCP"
gcloud config set project udmi
gcloud auth activate-service-account --key-file ../../cloud/gcp/auth/credentials.json
./deploy.sh