K
Q

Question

Circleci deployment to Google Container Engine

10/25/2017

I manage to push my image to google cloud registery but when I try to rolling upgrade my cluster nothing happen.

I check my pods and they are not restarted.

I think I'm missing a command to force the rolling upgrade but I can't found help into the circle ci documentation wich is really low : https://circleci.com/docs/2.0/google-container-engine/

my circle.yml :

version: 2

jobs:
   build:
     docker:
       - image: google/cloud-sdk:latest
     steps:

        - checkout

        - run:
            name: Install Docker client
            command: |
              set -x
              VER="17.03.0-ce"
              curl -L -o /tmp/docker-$VER.tgz https://get.docker.com/builds/Linux/x86_64/docker-$VER.tgz
              tar -xz -C /tmp -f /tmp/docker-$VER.tgz
              mv -f /tmp/docker/* /usr/bin

        - run:
            name: Install Docker Compose
            command: |
              set -x
              curl -L https://github.com/docker/compose/releases/download/1.11.2/docker-compose-`uname -s`-`uname -m` > /tmp/docker-compose
              mv -f /tmp/docker-compose /usr/local/bin/docker-compose
              chmod +x /usr/local/bin/docker-compose

        - type: setup-docker-engine

        - run:
            name: Build Docker image
            command: |
              docker build -t eu.gcr.io/${GOOGLE_PROJECT_ID}/gcp-test:latest .

        - run:
            name: Dump Google Cloud Credentials to file
            command: |
              echo ${GOOGLE_AUTH} > ${HOME}/gcp-key.json
              gcloud auth activate-service-account --key-file ${HOME}/gcp-key.json
              gcloud --quiet config set project ${GOOGLE_PROJECT_ID}
              gcloud --quiet config set compute/zone ${GOOGLE_COMPUTE_ZONE}
              gcloud --quiet container clusters get-credentials ${GOOGLE_CLUSTER_NAME}

        - run:
            name: Push the image to google registery
            command: |
              gcloud docker -- push eu.gcr.io/${GOOGLE_PROJECT_ID}/gcp-test:latest

        - run:
            name: Deploy image to the gc cluster
            command: |
              kubectl apply -f deployment.yml
              kubectl apply -f service.yml
              kubectl set image deployment/solive-deployment node-app=eu.gcr.io/${GOOGLE_PROJECT_ID}/gcp-test:latest
              kubectl get deployments
              kubectl get pods

-- Stephane Karagulmez

circleci

google-kubernetes-engine

2 Answers

2/18/2018

A solution to force the re-pull of the image is to change the pod-template hash during each build. This can be achieved by adding an environment variable that is altered during build:

deployment.yaml

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: demo
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: demo
    spec:
      containers:
      - name: demo
        image: registry.example.com/apps/demo:master
        imagePullPolicy: Always
        env:
        - name: FOR_GODS_SAKE_PLEASE_REDEPLOY
          value: 'THIS_STRING_IS_REPLACED_DURING_BUILD'

deploy:

sed -ie "s/THIS_STRING_IS_REPLACED_DURING_BUILD/$(date)/g" deployment.yml
kubectl apply -f deployment.yml

Full credits to the original answer - https://github.com/kubernetes/kubernetes/issues/33664#issuecomment-292895327

-- user1502

Source: StackOverflow

12/6/2017

Pods do not restart when you do kubectl apply -f deployment.yml, but the image they are running on will be updated (the container will be stopped and started with the new image).

In my experience it is not necessary to update the service or to set image.

Note that my answer shouldn't be taken for granted as I am far from an expert in Kubernetes.

-- Thomas Sauvajon

Source: StackOverflow

KQ

Circleci deployment to Google Container Engine

Similar Questions

2 Answers

K
Q