Packer won't create image on GCloud

Question

I've been using the tools from HashiCorp (Packer and Terraform) to create images and then build VMs by using Jenkins on AWS. Now I have a project that is on GCloud and I have successfully configured Terraform to create a VM but when I try to run packer I get the following error root@packer packer]# packer build createImage.json

googlecompute output will be in this color.

==> googlecompute: Checking image does not exist...
==> googlecompute: Creating temporary SSH key for instance...
==> googlecompute: Error getting source image for instance creation: Could not find image, centos-7, in projects, [<PROJECT_ID> centos-cloud coreos-cloud debian-cloud google-containers opensuse-cloud rhel-cloud suse-cloud ubuntu-os-cloud windows-cloud gce-nvme]: 11 error(s) occurred:
==> googlecompute: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body: 
==> googlecompute: * googleapi: got HTTP response code 400 with body:
Build 'googlecompute' errored: Error getting source image for instance creation: Could not find image, centos-7, in projects, [PROJECT_ID centos-cloud coreos-cloud debian-cloud google-containers opensuse-cloud rhel-cloud suse-cloud ubuntu-os-cloud windows-cloud gce-nvme]: 11 error(s) occurred:

* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 

==> Some builds didn't complete successfully and had errors:
--> googlecompute: Error getting source image for instance creation: Could not find image, centos-7, in projects, [PROJECT_ID centos-cloud coreos-cloud debian-cloud google-containers opensuse-cloud rhel-cloud suse-cloud ubuntu-os-cloud windows-cloud gce-nvme]: 11 error(s) occurred:

* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 
* googleapi: got HTTP response code 400 with body: 

==> Builds finished but no artifacts were created.

This is my createImage.json

{
 "variables": {
    "project_id": "PROJECT_ID"
  },
 "builders": [
    {
      "type": "googlecompute",
      "region": "europe-west1",
      "zone": "europe-west1-b",
      "project_id": "{{user `project_id`}}",
      "source_image_family": "centos-7",
      "image_name": "My-Test-Image",
      "disk_size": 10,
      "machine_type": "f1-micro",
      "ssh_username": "centos"
    }
  ]
}

I should mention that the service account that I am using is even given a owner permission just to be sure that I have no issues with the permissions. Does anybody has any idea why I have this issue while on AWS it works great.

Answer 1

As you're not using "account_file" to specify a JSON Service Account key, I assume that you're running packer from a VM instance in GCP that is running as a Service Account, and that such service account is the one with Owner role.

I would first check that the service account is properly configured in the VM details and that the Cloud API Access Scopes are defined as described in this Packer doc:

• "Read Write" for Compute Engine
• "Full" for "Storage"

Additionally, even though I think it's not related to this issue, you shouldn't build images using f1-micro nor g1-small machine types as per this doc.

Finally, I would suggest to post this question also in the Packer Community Forum or any other of the official channels.

Answer 2

Can you list the available images on GCloud and use this information for your json configuration file:

gcloud compute images list --filter centos