Image build takes a very long time

[robertd77]

Hi, not sure it this an issue or expected - when creating a custom Windows image with the command, the build takes a very long time, over 40 minutes. Just wanted to know if this is normal and if not why this might be taking so long. The command output shows 'Starting Run template...' and then takes about 40 minutes to succeed. Here is my command being run from the Github Actions worflow:

• name: Create VM Image
  id: imageBuilder
  uses: azure/build-vm-image@v0
  with:
  location: "eastus"
  resource-group-name: "test-group"
  managed-identity: "Mission"
  vm-size: "Standard_B2s"
  source-os-type: "windows"
  source-image-type: "platformImage"
  source-image: MicrosoftWindowsServer:WindowsServer:2019-Datacenter:latest
  customizer-source: "${{ GITHUB.WORKSPACE }}/dist/angular-test-app"
  dist-resource-id: "/subscriptions/${{ secrets.SUBSCRIPTION_ID }}/resourceGroups/test-group/providers/Microsoft.Compute/galleries/testGallery/images/testVM"
  dist-type: "SharedImageGallery"
  dist-location: "eastus"

[github-actions]

This issue is idle because it has been open for 14 days with no activity.

[jhitze]

Hi @olayemio! I saw that this is assigned to you. Do you have an idea of where the issue may be? I'd love to help speed it up.

[github-actions]

This issue is idle because it has been open for 14 days with no activity.

[tikicoder]

After dealing with this so much I think I can break it down partially.
If you watch the image builder spins up a set of resources including a load balancer, key vault, and storage account, I want to say around 10 to 15 ish also depending on your configuration. So of them feel like a waste since you cannot really use them, like the key vault or load balancer. So that would be some wasted time.

Then the other large issue is the distribution. We are distributing it to 2 regions, the local and then 1 other. I have watched it take 20 to 30 minutes to do the 2 regions.

So even if the building and setup take 20 minutes you are still waiting for the distribution. I have a PR where I look to add the ability to set a run variable and only have the pipeline wait until the distro starts. If the distro fails The only way to fix it would be to rerun the pipeline, but at that point, you know the build process is good, so to me from a GitHub action standpoint it could end. If I was going to build something that would use that image to make another, I would know to have it wait an hour before running and then it would be good. I could even add logic to check.

PR in question
#17

The other issue is that this action just seems like it's left out. I am tempted to fork it just so I can apply the work I have done to it.

[github-actions]

This issue is idle because it has been open for 14 days with no activity.