How set up bitbucket pipeline using a gcp private image via OIDC?

[lucas_ferreira]

Hi, I am new into this, but I am trying to set up a pipeline using a docker image that is on google registry. I am having a some problems doing this and I can't find a good documentation. Can anyone help me? I already set a google identity federation and a pool on gcp.

pipelines:

     default:

      - step:

            oidc: true

            name: Build 

            image: gcr.io/myproject/it@sha2:123

      - script:

                ....  I am stucked, I don't know if I have to use de UUID here or the audience url and how to do it.

[Justin Thomas · Accepted solution]

@lucas_ferreira Bitbucket Pipelines currently doesn't support OpenID Connect with Google container registry, it's only supported with AWS. Currently, you can only use OpenID Connect token to integrate with GCP from within the step.

[Michal Czeraszkiewicz]

@Justin Thomas could You share the simplest example how to use Bitbucket OIDC with GCP?

Currently I have:

image: atlassian/default-image:3

pipelines:
  default:
  - parallel:
    - step: &docker-build-push
      name: Build and push images to GCR
      oidc: true
      image: google/cloud-sdk:alpine
      script:
      - echo "${BITBUCKET_STEP_OIDC_TOKEN}" > /tmp/credential-source-file.out
      - gcloud iam workload-identity-pools create-cred-config projects/${PROJECT_ID}/locations/global/workloadIdentityPools/bitbucket-pipelines/providers/bitbucket-pipelines --service-account="name@${PROJECT_ID}.iam.gserviceaccount.com" --output-file=/tmp/FILEPATH.json --credential-source-file=/tmp/credential-source-file.out --credential-source-type=text
      - gcloud auth login --cred-file=/tmp/FILEPATH.json
      - CLOUDSDK_CORE_DISABLE_PROMPTS=1 gcloud components install alpha
      - gcloud --project ${PROJECT_ID} alpha storage ls

But I get an error:

google.auth.exceptions.OAuthError: ('Error code invalid_target: The target service indicated by the "audience" parameters is invalid. This might either be because the pool or provider is disabled or deleted or because it doesn\'t exist.', '{"error":"invalid_target","error_description":"The target service indicated by the \\"audience\\" parameters is invalid. This might either be because the pool or provider is disabled or deleted or because it doesn\'t exist."}')

[Justin Thomas]

Hi @Michal Czeraszkiewicz , Following is an example to integrate with GCP using OpenID Connect. Please let me know if you have any further questions.

Step 1: Creating a Workload Identity Pool

In this step, you are going to create a workload identity pool which is a container for a collection of external identities.

gcloud beta iam workload-identity-pools create bitbucket-pipelines-oidc-demo \
--location="global" \
--description="A workload identity pool for demoing oidc integration" \
--display-name="bitbucket-pipelines-oidc-demo"

Step 2: Add OIDC Provider to The Workload Identity Pool

In this step, you are going to configure Bitbucket Pipelines OIDC provider as an OIDC provider for the workload identity pool created in the last step.

gcloud beta iam workload-identity-pools providers update-oidc bitbucket-oidc-idp \
--workload-identity-pool="bitbucket-pipelines-oidc-demo" \
--issuer-uri="https://api.bitbucket.org/2.0/workspaces/{WORKSPACE}/pipelines-config/identity/oidc" \
--location="global" \
--attribute-mapping="google.subject=assertion.sub,attribute.workspace_uuid=assertion.workspaceUuid" \
--allowed-audiences="ari:cloud:bitbucket::workspace/{WORKSPACE_UUID}"

Note: workspace_uuid for audience is without curly brackets.

Step 3: Create a GCP Service Account

In the step, you are going to create a GCP service account, since external identities can't access most GCP resources directly.

gcloud iam service-accounts create bitbucket-pipelines-oidc-demo \
--display-name="bitbucket-pipelines-oidc-demo"

Note: This step can be done from GCP console.

Step 4: Allowing Bitbucket Pipelines Step to Impersonate Service Account

In this step, you are going to allow Bitbucket Pipelines Step (Using generated ID Token) to impersonate a service account.

gcloud iam service-accounts add-iam-policy-binding {SERVICE_ACCOUNT_EMAIL} \
--role="roles/iam.workloadIdentityUser" \
--member="principalSet://iam.googleapis.com/projects/{GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/bitbucket-pipelines-oidc-demo/attribute.workspace_uuid/{WORKSPACE_UUID}"

Step 5: Configure Build to Exchange OIDC ID Token for a Google Token

In the step, you are going to configure your build to exchange the OIDC ID token for a google token that can be used to access GCP resources.

Note: The bitbucket-pipelines.yml is using a shell script file to get the access token since this feature is a beta version and gcloud doesn’t have commands for doing the required steps.

bitbucket-pipelines.yml

image: google/cloud-sdk
pipelines:
   default:
       - step:
           oidc: true
           script:
               - ./getGoogleAccessToken.sh
               - export GCP_ACCESS_TOKEN=$(jq '.accessToken' accessToken.json)

getGoogleAccessToken.sh

#!/bin/bash

apt-get install -qq jq
sed -i "s/OIDC_TOKEN/$BITBUCKET_STEP_OIDC_TOKEN/g" getFederatedToken.json
curl -X POST \
     -H "Content-Type: application/json; charset=utf-8" \
     -d @getFederatedToken.json \ 
     https://sts.googleapis.com/v1beta/token > federatedToken.json
FEDERATED_TOKEN=$(jq '.access_token' federatedToken.json)
curl -X POST \
     -H "Authorization: Bearer $FEDERATED_TOKEN" \
     -H "Content-Type: application/json; charset=utf-8" \
     -d '{"scope": ["https://www.googleapis.com/auth/cloud-platform"]}' \
     https://iamcredentials.googleapis.com/v1/projects/-/serviceAccounts/{SERVICE_ACCOUNT_EMAIL}:generateAccessToken > accessToken.json

getFederatedToken.json

{
"audience": "//iam.googleapis.com/projects/{GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/bitbucket-pipelines-oidc-demo/providers/bitbucket-oidc-idp",
"grantType": "urn:ietf:params:oauth:grant-type:token-exchange",
"requestedTokenType": "urn:ietf:params:oauth:token-type:access_token",
"scope": "https://www.googleapis.com/auth/cloud-platform",
"subjectTokenType": "urn:ietf:params:oauth:token-type:jwt",
"subjectToken": "OIDC_TOKEN"
}

[Michal Czeraszkiewicz]

@Justin Thomas thanks a lot for the very helpful support!

I ended up with this setup (for testing GCP auth):

image: atlassian/default-image:3

pipelines:
  default:
  - parallel:
    - step: &docker-build-push
      name: Build and push images to GCR
      oidc: true
      image: google/cloud-sdk:alpine
      script:
      - gcloud --version
      - export GCP_PROJECT_NUMBER=12345678901
      - export PROJECT_NAME=my-project-name
      - export workload_identity_pool_name='bitbucket-pipelines'
      - export workload_identity_pool_provider_name='bitbucket-pipelines'
      - export SERVICE_ACCOUNT_EMAIL="sa-name@${PROJECT_NAME}.iam.gserviceaccount.com"
      - echo -n "${BITBUCKET_STEP_OIDC_TOKEN}" > /tmp/gcp_access_token.out
      - gcloud iam workload-identity-pools create-cred-config "projects/${GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/${workload_identity_pool_name}/providers/${workload_identity_pool_provider_name}" --credential-source-file=/tmp/gcp_access_token.out --service-account="${SERVICE_ACCOUNT_EMAIL}" --output-file=sts-creds.json
      - export GOOGLE_APPLICATION_CREDENTIALS=`pwd`/sts-creds.json
      - gcloud auth login --cred-file=`pwd`/sts-creds.json
      - gcloud info
      - CLOUDSDK_CORE_DISABLE_PROMPTS=1 gcloud components install alpha
      - gcloud --project "${PROJECT_NAME}" alpha storage ls gs://some-bucket/

I wish this (+ what You posted) would be in the official Atlassian documentation :-|. This would save me a lot of time.

Below my feedback to the commands You posted because it was NOT clear for me when to use the mustache brackets:

export WORKSPACE_NAME='my-workspace'
export WORKSPACE_UUID='bebebebe-bebe-bebe-bebe-bebebebebebe'

# no mustache brackets
gcloud beta iam workload-identity-pools providers update-oidc bitbucket-oidc-idp \
  --workload-identity-pool="bitbucket-pipelines-oidc-demo" \
  --issuer-uri="https://api.bitbucket.org/2.0/workspaces/${WORKSPACE_NAME}/pipelines-config/identity/oidc" \
  --location="global" \
  --attribute-mapping="google.subject=assertion.sub,attribute.workspace_uuid=assertion.workspaceUuid" \
  --allowed-audiences="ari:cloud:bitbucket::workspace/${WORKSPACE_UUID}"

export GCP_PROJECT_NUMBER='12345678901'
export PROJECT_NAME='my-project-name'
export SERVICE_ACCOUNT_EMAIL="sa-name@${PROJECT_NAME}.iam.gserviceaccount.com"
export WORKSPACE_UUID='bebebebe-bebe-bebe-bebe-bebebebebebe'
gcloud iam service-accounts add-iam-policy-binding "${SERVICE_ACCOUNT_EMAIL}" \
  --role="roles/iam.workloadIdentityUser" \
  --member="principalSet://iam.googleapis.com/projects//locations/global/workloadIdentityPools/bitbucket-pipelines-oidc-demo/attribute.workspace_uuid/{${WORKSPACE_UUID}}"

Once again thanks for the fast and very helpful feedback!

[Michal Czeraszkiewicz]

@Justin Thomas thank You for posting the useful instructions!

I ended up with a little bit simpler version:

image: atlassian/default-image:3

pipelines:
  default:
  - parallel:
    - step: &docker-build-push
      name: Build and push images to GCR
      oidc: true
      image: google/cloud-sdk:alpine
      script:
      - export GCP_PROJECT_NUMBER=12345678901
      - export PROJECT_NAME=my-project-name

      - export workload_identity_pool_id='bitbucket-pipelines-oidc-demo'
      - export workload_identity_pool_provider_id='update-oidc bitbucket-oidc-idp'
      - export SERVICE_ACCOUNT_EMAIL="sa-name@${PROJECT_NAME}.iam.gserviceaccount.com"

      - echo -n "${BITBUCKET_STEP_OIDC_TOKEN}" > /tmp/gcp_access_token.out

      - gcloud iam workload-identity-pools create-cred-config "projects/${GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/${workload_identity_pool_id}/providers/${workload_identity_pool_provider_id}" --credential-source-file=/tmp/gcp_access_token.out --service-account="${SERVICE_ACCOUNT_EMAIL}" --output-file=sts-creds.json

      - export GOOGLE_APPLICATION_CREDENTIALS=`pwd`/sts-creds.json
      - gcloud auth login --cred-file=`pwd`/sts-creds.json
      - CLOUDSDK_CORE_DISABLE_PROMPTS=1 gcloud components install alpha
      - gcloud --project "${PROJECT_NAME}" alpha storage ls gs://some-bucket/

Below feedback to the commands which You posted - I found the usage of mustache brackets very confusing :-(...

export WORKSPACE_NAME='my-workspace'
export WORKSPACE_UUID='bebebebe-bebe-bebe-bebe-bebebebebebe'
gcloud beta iam workload-identity-pools providers update-oidc bitbucket-oidc-idp \
  --workload-identity-pool="bitbucket-pipelines-oidc-demo" \
  --issuer-uri="https://api.bitbucket.org/2.0/workspaces/${WORKSPACE_NAME}/pipelines-config/identity/oidc" \
  --location="global" \
  --attribute-mapping="google.subject=assertion.sub,attribute.workspace_uuid=assertion.workspaceUuid" \
  --allowed-audiences="ari:cloud:bitbucket::workspace/${WORKSPACE_UUID}"

export GCP_PROJECT_NUMBER='12345678901'
export PROJECT_NAME='my-project-name'
export SERVICE_ACCOUNT_EMAIL="sa-name@${PROJECT_NAME}.iam.gserviceaccount.com"
export WORKSPACE_UUID='bebebebe-bebe-bebe-bebe-bebebebebebe'

gcloud iam service-accounts add-iam-policy-binding "${SERVICE_ACCOUNT_EMAIL}" \
  --role="roles/iam.workloadIdentityUser" \
  --member="principalSet://iam.googleapis.com/projects//locations/global/workloadIdentityPools/bitbucket-pipelines-oidc-demo/attribute.workspace_uuid/{${WORKSPACE_UUID}}"

I wish this + what You posted would be in the official Atlassian documentation. This would save me a lot of time.

Once again BIG thanks for the fast and very useful feedback

[Michal Czeraszkiewicz]

@Justin Thomas thank You for posting the useful instructions!

I ended up with a little bit simpler version:

image: atlassian/default-image:3

pipelines:
  default:
  - parallel:
    - step: &docker-build-push
      name: Build and push images to GCR
      oidc: true
      image: google/cloud-sdk:alpine
      script:
      - export GCP_PROJECT_NUMBER=12345678901
      - export PROJECT_NAME=my-project-name

      - export workload_identity_pool_id='bitbucket-pipelines-oidc-demo'
      - export workload_identity_pool_provider_id='update-oidc bitbucket-oidc-idp'
      - export SERVICE_ACCOUNT_EMAIL="sa-name@${PROJECT_NAME}.iam.gserviceaccount.com"

      - echo -n "${BITBUCKET_STEP_OIDC_TOKEN}" > /tmp/gcp_access_token.out

      - gcloud iam workload-identity-pools create-cred-config "projects/${GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/${workload_identity_pool_id}/providers/${workload_identity_pool_provider_id}" --credential-source-file=/tmp/gcp_access_token.out --service-account="${SERVICE_ACCOUNT_EMAIL}" --output-file=sts-creds.json

      - export GOOGLE_APPLICATION_CREDENTIALS=`pwd`/sts-creds.json
      - gcloud auth login --cred-file=`pwd`/sts-creds.json
      - CLOUDSDK_CORE_DISABLE_PROMPTS=1 gcloud components install alpha
      - gcloud --project "${PROJECT_NAME}" alpha storage ls gs://some-bucket/

Below feedback to the commands which You posted - I found the usage of mustache brackets very confusing :-(...

export WORKSPACE_NAME='my-workspace'
export WORKSPACE_UUID='bebebebe-bebe-bebe-bebe-bebebebebebe'
gcloud beta iam workload-identity-pools providers update-oidc bitbucket-oidc-idp \
  --workload-identity-pool="bitbucket-pipelines-oidc-demo" \
  --issuer-uri="https://api.bitbucket.org/2.0/workspaces/${WORKSPACE_NAME}/pipelines-config/identity/oidc" \
  --location="global" \
  --attribute-mapping="google.subject=assertion.sub,attribute.workspace_uuid=assertion.workspaceUuid" \
  --allowed-audiences="ari:cloud:bitbucket::workspace/${WORKSPACE_UUID}"

export GCP_PROJECT_NUMBER='12345678901'
export PROJECT_NAME='my-project-name'
export SERVICE_ACCOUNT_EMAIL="sa-name@${PROJECT_NAME}.iam.gserviceaccount.com"
export WORKSPACE_UUID='bebebebe-bebe-bebe-bebe-bebebebebebe'

gcloud iam service-accounts add-iam-policy-binding "${SERVICE_ACCOUNT_EMAIL}" \
  --role="roles/iam.workloadIdentityUser" \
  --member="principalSet://iam.googleapis.com/projects//locations/global/workloadIdentityPools/bitbucket-pipelines-oidc-demo/attribute.workspace_uuid/{${WORKSPACE_UUID}}"

I wish this + what You posted would be in the official Atlassian documentation. This would save me a lot of time.

Once again BIG thanks for the fast and very useful feedback!

[Justin Thomas]

@Michal Czeraszkiewicz It's great to hear it worked. Yes, we will be adding these steps to the documentation.

[Michal Czeraszkiewicz]

Happy to hear that this will land in the documentation!

Another feedback point is: for some unclear reason my reply was disappearing after I posted it.

I understand that this might be part of the moderation process... But it would be nice to get some feedback about this ;-) Otherwise I'm confused and frustrated that the reply I just wrote is gone!