无法从GCP Marketplace部署Migrate for Compute Engine（Velostrata）求助

I’ve helped my team work through similar Velostrata deployment hiccups before, so let’s break down targeted fixes for this 412 precondition failure error you’re hitting:

• Double-check your Velostrata service account’s role assignments
  Even though you’ve created the service account via script and manually, make sure it’s assigned the exact roles Velostrata requires—your own admin permissions don’t automatically pass to the service account. Critical roles include velostrata.serviceAgent plus Compute Engine-specific permissions. Verify this with the gcloud CLI:

  gcloud projects get-iam-policy YOUR_PROJECT_ID --filter="bindings.members:serviceAccount:YOUR_VELOSTRATA_SA@YOUR_PROJECT_ID.iam.gserviceaccount.com"
  

  Also, ensure no conditional IAM restrictions (like region locks or time-based limits) are applied to the service account—these can trigger precondition failures out of nowhere.

• Confirm dependent APIs are fully enabled
  412 errors often pop up when required services aren’t turned on. Make sure these APIs are enabled in your project:

  • Compute Engine API (compute.googleapis.com)
  • Velostrata API (velostrata.googleapis.com)
  • Cloud Resource Manager API (cloudresourcemanager.googleapis.com)
    You can check via the Cloud Console API Library, or run this command to list enabled services:

  gcloud services list --enabled | grep -E "(compute|velostrata|cloudresourcemanager)"
  

• Check organization-level policies and constraints
  Even with Organization Admin and Project Owner roles, your org might have policies blocking Velostrata’s instance creation. Look for constraints that restrict instance deployment or service account usage:

  gcloud resource-manager org-policies list --organization=YOUR_ORG_ID
  

  Pay extra attention to policies like constraints/compute.restrictInstanceCreation that could block the GCE instances Velostrata needs to spin up.

• Validate project quotas and resource availability
  The error points to a compute.v1.instance precondition failure, which might mean your project lacks sufficient quotas for the resources Velostrata requires. Check your quota limits in the Cloud Console (IAM & Admin > Quotas) for CPU, memory, and external IP addresses in your target region. Also confirm the region/zone you’re deploying to has no ongoing resource shortages.

• Reset the deployment state
  Temporary cache or stale configuration can sometimes cause issues. Try these steps to start fresh:

  1. Delete any existing Velostrata service accounts in your project.
  2. Re-run the official setup script to create a new service account:

     curl -O https://raw.githubusercontent.com/GoogleCloudPlatform/velostrata-tools/master/deployment/setup_service_account.py
     python3 setup_service_account.py --project YOUR_PROJECT_ID
     

  3. Clear your browser cache or use an incognito window to re-initiate the Marketplace deployment.

If none of these steps fix the issue, dive into detailed logs in the Cloud Console’s Log Explorer. Filter for resource.type="gce_instance" and severity="ERROR"—the specific log entries should spell out exactly which precondition is failing, making it easier to resolve.

内容的提问来源于stack exchange，提问作者Ravi k