azure / k8s-actions Goto Github PK

Hello,

where is the directory I should put this manifests?

I'm receiving this error:

#[error]Error: ENOENT: no such file or directory, open 'manifests/deployment.yml'

Typo in action inputs

In the aks-set-context command it's not clear that activeDirectoryEndpointUrl and resourceManagerEndpointUrl are required values in the Azure credentials, and the error message doesn't mention them.

I spent some time debugging this using an existing service principal which didn't have the same output as the one in the docs, and it might be more straightforward to either add those values to the error message or set defaults of "https://login.microsoftonline.com" and "https://management.azure.com/" if those will almost always be the values.

I'm happy to put in a PR for this if it would be helpful, and thanks for maintaining this repo! It was really helpful in getting started deploying to AKS with actions

Hello!

I was having some issues getting setup-kubectl working properly.

I kept getting:

##[warning]GetStableVersionFailed
##[error]Error: DownloadKubectlFailed

when trying to use this basic incantation for it: https://github.com/djquan/k8s-action-repro/blob/master/.github/workflows/main.yml#L18-L20

I tracked down the issue to the use of an out-of-date tool-cache that was using the wrong environment variables. Specifically, the version of tool-cache that this repository is using is looking for RUNNER_TEMPDIRECTORY and RUNNER_TOOLSDIRECTORY, when the variables in the actual runner are RUNNER_TEMP and RUNNER_TOOL_CACHE. This was updated in tool-cache recently, actions/toolkit@5218a83#diff-ecd700b4a481c69908b4b61fc89806ad.

This reproduction case should show the correct variables the runner has, the method I eventually used to get setup-ctl working, and the error it fails with if the old environment variables are not set. The results from the checks are here.

Apologies if I missed anywhere this behavior was documented. I did try to update the packages on a fork, but didn't get too far before running into some issues that I couldn't get past.

Reproduction case:

name: Build
on:
  push:
    branches:
      - "master"
jobs:
  build:
    runs-on: ubuntu-latest
    steps:
    - name: display env
      run: env | grep RUNNER
    - name: azure with env var
      uses: azure/k8s-actions/setup-kubectl@master
      continue-on-error: true
      env:
        RUNNER_TEMPDIRECTORY: ${RUNNER_TEMP}
        RUNNER_TOOLSDIRECTORY: ${RUNNER_TOOL_CACHE}
    - name: azure without env var
      uses: azure/k8s-actions/setup-kubectl@master
      continue-on-error: true

I have the following problem:
When i add a secret to an existing namespace in the cluster: /usr/bin/kubectl create secret generic mysql-secret --from-literal=MYSQL_DATABASE=*** --from-literal=MYSQL_USERNAME=*** -- from-literal=MYSQL_PASSWORD=*** -n magento-test-develop

It adds the secret to the default namespace but the custom namespace is there. And the expected behaviour should be that it adds the secret to the namespace given.

Is k8s-set-secret, should be k8s-create-secret

When trying to update the image reference in my deployment script, there's an issue.

In my action file I have

        - uses: azure/k8s-actions/k8s-deploy@master
        with:
          manifests: |
            deployment.yml
          images: |
            my.repo.com:1234/helloworld:${{ github.sha }}

When I check the "k8s-deploy/src/kubernetes-utils.ts", I can see the to find which image should be replace there's a

let imageName = container.split(':')[0];

Which can't work in this case. I think it should be:

let imageName = container;
if (imageName.indexOf(':') > 0) {
	imageName = imageName.substring(0, imageName.lastIndexOf(':'));
}

All k8s actions have some logics which are duplicate.

• install kubectl logic
• common util functions i.e. get runner temp directory logic