K
Q

kubectl set image deployment doesn't set new container image (gcp)

10/26/2019
kubectl set image deployment/$DEPLOYMENT_EXTENSION $INSTANCE_NAME=gcr.io/$PROJECT_ID/$DOCKER_REPOSITORY:latest

I use this command do load new created image to my existing cluster (update version of my app) . But when I do it , and then go to the site , I don't see any changes .

spec:
      terminationGracePeriodSeconds: 30
      containers:
      - name: booknotes
        image: gcr.io/my-image:latest
        imagePullPolicy: Always

I've also added this 2 lines in deployment.yaml file and applied it for my cluser:

imagePullPolicy: Always
terminationGracePeriodSeconds: 30

But it still doesn't work. Can it be because I use :latest tag ? Or it isn't related ? If you have some idias pls let me know. And also if you need additional info , I will attach it !

gitlab-ci.yml

   stages:
      - build
      - docker-push
      - deploy

    cache:
      paths:
        - node_modules/

    build:
      stage: build
      image: node:latest
      script:
        - yarn install
        - npm run build
      artifacts:
        paths:
          - dist/
      only:
        - master

    docker:
      stage: docker-push
      image: docker:18.09.7

      services:
        - docker:18.09.7-dind
        - google/cloud-sdk:latest

      script:
        - echo $GCP_ACCESS_JSON > $CI_PIPELINE_ID.json
        - cat $CI_PIPELINE_ID.json | docker login -u _json_key --password-stdin $GCP_REGION
        - docker build -t gcr.io/$PROJECT_ID/$DOCKER_REPOSITORY:latest .
        - docker push gcr.io/$PROJECT_ID/$DOCKER_REPOSITORY:latest

      only:
        - master

    test:
      stage: deploy
      image: google/cloud-sdk:latest
      script:
        - echo $GCP_ACCESS_JSON > $CI_PIPELINE_ID.json
        - gcloud auth activate-service-account $GCP_CE_PROJECT_EMAIL --key-file $CI_PIPELINE_ID.json --project $GCP_PROJECT_ID
        - gcloud container clusters get-credentials $CLUSTER_NAME --zone $ZONE --project $PROJECT_NAME
        - kubectl set image deployment/$DEPLOYMENT_EXTENSION $INSTANCE_NAME=gcr.io/$PROJECT_ID/$DOCKER_REPOSITORY:latest

      only:
        - master
-- Andrey Radkevich
gitlab-ci
google-cloud-platform
kubernetes

Similar Questions

How to broadcast gRPC streams inter pods
How to create and mount common data for all pods
Deploy Nodejs app on google kubernetes engine using Docker and PM2
kubectl replace -f says it has replaces a resource, but it actually hasn't
Is it possible to install curl into busybox in kubernetes pod
k8s 1.18.1: api not reachable since update to 1.18.1
pulling image from private docker registry causes filesystem layer verification failed for digest sha256:

2 Answers

10/26/2019

You are correct, the latest tag will not trigger a re-deployment of your pod. Instead you could set the image and then scale the deployment.

kubectl set image deployment/alpine alpine=alpine:latest
kubectl scale deployment --replicas=0
kubectl scale deployment --replicas=1

Another option would be to set the image and then patch an annotation or change some yaml field.

-- Rodrigo Loza
Source: StackOverflow
10/26/2019

This question, the symptoms and the reasons are very close to [this one],(Google cloud platform creating a pipeline with Kubernetes and replacing the same container)

Apply the same solution in your Gitlab-CI pipeline (use global variable to change the image tag and to deploy everytime a new one to force Kubernetes to pull and to deploy it).

I can help you on Gitlib-CI if you have difficulties to do this.

-- guillaume blaquiere
Source: StackOverflow