Pass through env vars into orb

Ecosystem Orbs
1

Not really sure if I haven’t got the syntax right or if what I’m trying to do is impossible. It seems like it should be doable but I haven’t been able to find an example that demonstrates what I’m trying to do and I keep getting config errors in my attempts.

version: 2.1

orbs:
  gcp-cli: circleci/gcp-cli@1.0

jobs:
  build_and_push:
    executor: gcp-cli/google
    environment:
    - PROJECT_NAME: "project"
    - GOOGLE_PROJECT_ID: "project-uid"
    - GOOGLE_COMPUTE_ZONE: "europe-west1-a"
    - GOOGLE_CLUSTER_NAME: "project-cluster"
    steps:
    - checkout
    - gcp-cli/initialize:
    - setup_remote_docker
    - run:
        name: Docker build and push
        command: |
          docker build \
            -t eu.gcr.io/${GOOGLE_PROJECT_ID}/${PROJECT_NAME}:${CIRCLE_SHA1} .
          gcloud auth print-access-token | docker login -u oauth2accesstoken --password-stdin https://eu.gcr.io
          docker push eu.gcr.io/${GOOGLE_PROJECT_ID}/${PROJECT_NAME}:${CIRCLE_SHA1}

So I’ve looked at the gcp-cli orb docs and I can see a lot of the default values are like default: $GOOGLE_PROJECT_ID. So I’m assuming if I define these defaults along with other environment vars I need they’ll be parsed correctly.
I want a setup like above but the outcome I’m getting is:

# ERROR IN CONFIG FILE:
# jobs: build_and_push: 0 subschemas matched instead of one
#   jobs: build_and_push: steps: 1: 0 subschemas matched instead of one
#     jobs: build_and_push: steps: 1: expected type: String, found: Mapping
#     jobs: build_and_push: steps: 1: minimum size: [1], found: [0]
#   jobs: build_and_push: expected type: String, found: Mapping

I find this error message a little hard to read as it doesn’t represent very well the config I’ve typed or which line the error has occurred.

Is what I’m trying to do possible and is my syntax just bad in this case or will I need to duplicate environment declarations and explicitly pass them in to the gcp-cli/initialize command?

As far as I’ve been able to work out from the docs yes they need to be explicitly passed like so:

executor: gcp-cli/google
    environment:
    - PROJECT_NAME: "project"
    - GOOGLE_PROJECT_ID: "project-uid"
    - GOOGLE_COMPUTE_ZONE: "europe-west1-a"
    - GOOGLE_CLUSTER_NAME: "project-cluster"
    steps:
    - checkout
    - gcp-cli/initialize:
          gcloud-service-key: $GCLOUD_SERVICE_KEY # actually this one doesn't
          google-project-id: $GOOGLE_PROJECT_ID
          google-compute-zone: $GOOGLE_COMPUTE_ZONE

I assume it’s because they’re present in the job but not made available in the executor container. Unlike the service key which is because it’s project wide. So I guess my real question becomes can we add env variables into executors like we would a normal docker container?

docker:
      - image: someimage
        environment:
          ENV: CI

I’m really only saving on two lines of text in this example but it would still be nice to be able to add variables to the container.

2

Hello @vimto,
This may just be how it displays in the browser but it appears your steps (and env vars) need to be indented.

original:

 steps:
- checkout
- gcp-cli/initialize:
- setup_remote_docker

indented:

 steps:
    - checkout
    - gcp-cli/initialize:
    - setup_remote_docker

And the syntax highlighting in the forum helped me find the next issue.

- gcp-cli/initialize: The : signifies a mapping, where on the next line you would input parameters. However your next line is for a new step, so the : can be removed unless adding more data.

Try this:

 steps:
    - checkout
    - gcp-cli/initialize
    - setup_remote_docker
3

Wow yes you are right it was literally the : I had left behind :roll_eyes:. The indenting was fine, that I think is just the browser.

I was too busy thinking it wasn’t possible to check for basic yaml errors and I think the error message was different for what I’d seen before for yaml mistakes. Anyway sorry and thanks for your time.

4

Any time! Glad we got that resolved.