wololo
Get access

Provisioning Pipeline

When you complete onboarding, the platform creates a dedicated GCP VM for your agent fleet. The pipeline runs in ~30 seconds and progresses through 7 stages.

Stages

Provisioning Pipeline — ~30 seconds end to endqueuedRequest receivedvm_createGCP VM provisionedworkspace_bootstrapDirs + git reposagent_configopenclaw.json writtengateway_startGateway process startedconnectivity_checkWebSocket pingcompleteFleet active0s~30sAny stage failure → tenant status: suspended · dashboard shows last failed stage · retry available

What gets created

  • GCP VM (e2-standard-2, your project, europe-west2-b)
  • Service account with scoped IAM roles
  • Firewall rule for gateway port
  • SSH key pair (ephemeral, single-use)
  • openclaw.json — agent configuration written to VM

What triggers a reprovision?

Reset your tenant status to pending via the dashboard and click Provision. The pipeline runs fresh — the old VM is deleted first if it exists.

Failure handling

If any stage fails, the tenant status flips to suspended. The VM may still be running (cloud-init may have partially succeeded). The dashboard shows the last failed stage and a retry option.

The platform enforces a 240-second route timeout. If the full pipeline exceeds this (rare), the request returns suspended — but the VM setup script continues running in the background. Wait 60s and refresh the dashboard.