alerts/google-vertex-ai/metadata.yaml

alert_policy_templates:
  - id: "429-out-of-capacity"
    display_name: "Vertex AI 429 Out of Capacity Error"
    description: "This alert fires when at least 1% of requests to a Vertex AI publisher model responded with a 429 out-of-capacity error. This happens when cumulative demand on the model from customers in your ML processing location exceeds available compute."
    version: 1
    related_integrations:
      - id: vertex_ai
        platform: GCP
  - id: "pt-utilization-80"
    display_name: "80% of Provisioned Throughput Quota Reached"
    description: "This alert will be triggered if the project reaches 80% of its existing Provisioned Throughput quota limit for any model with an active Provisioned Throughput order."
    version: 1
    related_integrations:
      - id: vertex_ai
        platform: GCP
  - id: "pt-utilization-90"
    display_name: "90% of Provisioned Throughput Quota Reached"
    description: "This alert will be triggered if the project reaches 90% of its existing Provisioned Throughput quota limit for any model with an active Provisioned Throughput order."
    version: 1
    related_integrations:
      - id: vertex_ai
        platform: GCP
  - id: "pt-utilization-100"
    display_name: "100% of Provisioned Throughput Quota Reached"
    description: "This alert will be triggered if the project reaches 100% of its existing Provisioned Throughput quota limit for any model with an active Provisioned Throughput order. When this happens, further traffic will be completed under the shared tier or fail depending on the user’s configuration. "
    version: 1
    related_integrations:
      - id: vertex_ai
        platform: GCP