Deploy image is failing

General
1

Working scenario:

i have prepared the image using following commands in docker desktop which you can see below and pushed to ACR and then successfully deployed that image using AKS

shyam@DESKTOP-OO7EVK2:~$ docker tag aksimage:latest acrrrr.azurecr.io/aksimage:latest
shyam@DESKTOP-OO7EVK2:~$ docker push acrrrr.azurecr.io/aksimage:latest
The push refers to repository [acrrrr.azurecr.io/aksimage]
1fe6f44321a3: Pushed
936898278d02: Pushed
a702764f6c9e: Pushed
2dee78b1264b: Pushed
06e3bc688725: Pushed
2a58e20c885d: Pushed
86becefdf252: Pushed
65c636ce09f2: Pushed
latest: digest: sha256:466e1fe0b6b9316a5c8c72c37e72934de9cb2ccf3f717841e448d59cbd06cc9f size: 1992

FYI: I used ubhuntu 22.4 LTS since my laptops windows 11 in my case

Below is not working scenario :

where as here I am using same commands but I am running from pipeline and used same commands you can see below and I am using ubhuntu latest as agent which you can see below, here also image gets pushed but when I deploy I am seeing following error when deploy it using AKS

variables:
  imageName: 'aksimage'
  dockerfilePath: 'Dockerfile'
  buildContext: '.'
  containerRegistry: 'acrrrr.azurecr.io'
  dockerRegistryServiceConnection: 'acrserviceconnection'
  tag: 'latest2'  # Unique tag per build

pool:
  vmImage: 'ubuntu-latest'

stages:
# ========================
# 🔧 BUILD STAGE
# ========================
- stage: Build
  displayName: Build and Push Docker Image
  jobs:
  - job: Build
    displayName: Docker Build, Tag, Push
    steps:
    - script: |
        echo "Building image: $(imageName):$(tag) with platform linux/amd64"
        docker build -t $(imageName):$(tag) -f $(dockerfilePath) $(buildContext)
        docker tag $(imageName):$(tag) $(containerRegistry)/$(imageName):$(tag)
        docker images
      displayName: 'Build and Tag Docker Image'

    - task: Docker@2
      displayName: 'Push Docker Image to ACR'
      inputs:
        containerRegistry: 'ACR'
        repository: '$(imageName)'
        command: 'push'I
        tags: '$(tag)'

I am seeing following events from POD :in non working scenario

    DownwardAPI:             true
QoS Class:                   BestEffort
Node-Selectors:              <none>
Tolerations:                 node.kubernetes.io/not-ready:NoExecute op=Exists for 300s
                             node.kubernetes.io/unreachable:NoExecute op=Exists for 300s
Events:
  Type     Reason     Age                From               Message
  ----     ------     ----               ----               -------
  Normal   Scheduled  95s                default-scheduler  Successfully assigned default/aks-agent-deployment-5bc5ddf9b5-2tfkw to aks-np-36667834-vmss000001
  Normal   Pulling    94s                kubelet            Pulling image "acrrrr.azurecr.io/aksimage:latest2"
  Normal   Pulled     75s                kubelet            Successfully pulled image "acrrrr.azurecr.io/aksimage:latest2" in 18.364s (18.365s including waiting). Image size: 262281640 bytes.
  Normal   Created    28s (x4 over 75s)  kubelet            Created container: azp-agent
  Normal   Started    28s (x4 over 75s)  kubelet            Started container azp-agent
  Normal   Pulled     28s (x3 over 72s)  kubelet            Container image "acrrrr.azurecr.io/aksimage:latest2" already present on machine
  Warning  BackOff    0s (x6 over 71s)   kubelet            Back-off restarting failed container azp-agent in pod aks-agent-deployment-5bc5ddf9b5-2tfkw_default(9844a9b0-f1b4-4762-95f1-f846f4

Why it is failing during deployment? even both cases got pushed successfully , deployment failed in second scenario. it is same dockerfile and the code is same.

Please suggest

2

The only part where docker is involved in your process chain, is not causing the problem:

  • Your Azure DevOps pipeline seems to build and push the image to your private ACR repo. This is the only part where docker is involved.

The problem is in the AKS specific part, which has nothing to do with docker:

  • Your AKS deployment is able to pull the image from your ACR repo, and start a pod from it.
  • Most likely your probs are not correct, or the application inside container does not start, so the probes never find the endpoint they need to verify the container process is reachable.

This seems like a topic for the Azure DevOps community.

3

What is meany by probes?

and why I could deploy using same azure pipeline for different version image then? (of course that image got pushed manully )

4

Let me repeat myself: none of the resulting problems is related to docker. This is a case for an Azure DevOps Community.