Kubernetes

This page describes how to deploy PM415 on Google Cloud using Kubernetes

This PM415 Kubernetes deployment is for demos only - there is no persistence and data loss will occur. For a production deploy you must change the MySQL and Elastic Search Deployment files to include persistent storage, increase the cluster sizes, and enable node-to-node communication in Elastic Search.

Install and configure docker + gcloud

Follow the instructions to install docker and gcloud.

Add Google Container Registry to your docker config:

gcloud auth configure-docker

Login to your Google account:

gcloud auth login

Build and push container image to GCR

Download the PM415 git repository and change directories

git clone https://github.com/mreider/pm415.git
cd pm415

Build a docker image, tag it, and push to Google Container Registry (GCR). Replace the tag with your own Google Cloud Project name according to the format documented here.

docker build --tag gcr.io/pm415-238921/pm415 .
docker push gcr.io/pm415-238921/pm415

Create a Kubernetes cluster

Login to Google Cloud and navigate to Kubernetes Engine

Choose Create cluster

Choose Standard cluster and accept all the defaults

Click Create

Create a kubectl config

Install kubectl using these instructions

Return to the Kubernetes Engine page in the Google Cloud Console and click edit icon.

When the edit screen appears - write down the Endpoint IP address.

Click Show credentials and copy the certificate to your clipboard In a terminal

Convert the certificate to base64 - the following works on a mac:

pbpaste | base64

Return to the cluster credential window and copy the admin password

Create a file named config.yml with the following contents:

kind: Config
apiVersion: v1

clusters:
- cluster:
    server: https://(paste the endpoint IP address)
    certificate-authority-data: (paste the base64 cert here)
 name: default

current-context: default
contexts:
- context:
    cluster: default
    namespace: default
    user: default
  name: default

users:
- name: default
  user:
    username: admin
    password: (paste the admin password here)

Test your config

Create the KUBECONFIG environment variable to reference your config:

export KUBECONFIG=~/config.yml 

Test your config

kubectl get node

This should return the list of Kubernetes nodes in your Google Cloud account.

NAME                                                STATUS    ROLES     AGE       VERSION
gke-standard-cluster-1-444   Ready     <none>    22m       v1.11.8-gke.6
gke-standard-cluster-1-445   Ready     <none>    23m       v1.11.8-gke.6
gke-standard-cluster-1-446   Ready     <none>    23m       v1.11.8-gke.6

Create Kubernetes secrets

PM415 deployments have a few different secrets to reference.

kubectl create secret generic mysql --from-literal=password=RADOM_PASSWORD
kubectl create secret generic sendgrid --from-literal=password=YOUR_PASSWORD
kubectl create secret generic appkey --from-literal=password=RANDOM_KEY

Create pods

Edit the file pm415.yaml and replace the following values with your information. If you do not change the DOMAIN_URL to a valid domain - the installation will fail CORS security rules.

          - name: DOMAIN_URL
            value: your-domain.com
          - name: SENDGRID_USERNAME
            value: your-sendgrid-account

Create pods from the PM415 directory

 kubectl create -f mysql.yaml
 kubectl create -f es.yaml
 kubectl create -f pm415.yaml

Check the status of your pods

Check all of your pods to see if any are failing. If you see evictions, you might be running out of resources.

kubectl get pods --all-namespaces

Use the describe command to see specific error messages for a pod

kubectl describe pod pm415-575765d57b-g2pkg

Expose pods as services

This will expose two internal ip's for MySQL and ElasticSearch, and a Load Balancer for PM415.

kubectl create -f mysql-service.yaml
kubectl create -f elastic-search-service.yaml
kubectl create -f pm415-service.yaml

Get IP address and configure DNS

Wait a few minutes and then run this command to get the PM415 load balancer IP address:

kubectl get svc -l app=pm415

This will return information like the following:

pm415     LoadBalancer   10.7.250.50   35.238.238.234   80:31851/TCP   1m

Use this IP address for your DNS A record. Make sure this is the same value as the DOMAIN_URL you set in the pm415.yml deployment configuration.