Project ID: (PII Removed by Staff)
Problem Description:
I am experiencing a critical infrastructure issue where the default Google-managed Cloud Run Service Agent (service-<PROJECT_NUMBER> at gcp-sa-run.iam.gserviceaccount.com) is not being created for my project. This is preventing me from deploying any Cloud Run service that requires a Serverless VPC Connector.
Key Information:
- Using the Resource Manager getAncestry API, we have programmatically confirmed this is a standalone project with no parent Organization. The issue is therefore not related to an Organization Policy.
- The Serverless VPC Connector (
flipps-walkthrough
) is healthy and in theREADY
state. - The Serverless VPC Access API (
vpcaccess.googleapis.com
) is enabled.
Chronological Troubleshooting Steps Performed:
- Initial Deployment Failure: The initial
gcloud run deploy
command with the--vpc-connector
flag failed. The container did not start, and logs were empty. - Simplified Deployment Success: We successfully deployed a simplified version of the service without the VPC connector. This proves the basic Cloud Run functionality is working.
- IAM Verification Failure: Despite the successful deployment, the Cloud Run Service Agent (
gcp-sa-run
) was still not visible in the IAM console (with “Include Google-provided role grants” checked). - Programmatic Confirmation: We ran
gcloud iam service-accounts list
and confirmed programmatically that thegcp-sa-run
agent does not exist. - Forced IAM Binding Failure: An attempt to force a role binding to the agent’s expected email address failed with
INVALID_ARGUMENT: Service account ... does not exist.
- API Reset Cycle: We disabled and then re-enabled the Cloud Run API (
run.googleapis.com
) for the project. This did not trigger the agent’s creation. - Alternate Agent Check: We confirmed that the older, legacy
serverless-robot-prod
agent also does not exist.
Conclusion:
All standard and advanced troubleshooting steps have failed to provision the necessary Google-managed Cloud Run service agent. This appears to be a persistent, backend provisioning failure specific to this project. Please advise on how to escalate this for manual intervention.