integrations/tensorflow-serving/documentation.yaml

exporter_type: included
app_name_short: TF Serving
app_name: TensorFlow Serving
app_site_name: {{app_name}}
app_site_url: https://github.com/tensorflow/serving
exporter_name: {{app_name_short}}
exporter_repo_url: https://www.tensorflow.org/tfx/serving/serving_config#monitoring_configuration
additional_prereq_info: |
  {{app_name_short}} exposes Prometheus-format metrics when the `--monitoring_config_file` flag
  is used to specify a file containing a MonitoringConfig protocol buffer.
  
  This is an example:
  ```
  prometheus_config {
    enable: true,
    path: "/monitoring/prometheus/metrics"
  }
  ```
config_mods: |
  # Copyright 2025 Google LLC
  #
  # Licensed under the Apache License, Version 2.0 (the "License");
  # you may not use this file except in compliance with the License.
  # You may obtain a copy of the License at
  #
  #      http://www.apache.org/licenses/LICENSE-2.0
  #
  # Unless required by applicable law or agreed to in writing, software
  # distributed under the License is distributed on an "AS IS" BASIS,
  # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  # See the License for the specific language governing permissions and
  # limitations under the License.

  apiVersion: apps/v1
  kind: Deployment
  metadata:
    name: tfserve-deployment
    labels:
      app: tfserve-server
  spec:
    selector:
      matchLabels:
        app: tfserve
    replicas: 1
    template:
      metadata:
        labels:
          app: tfserve
        annotations:
          gke-gcsfuse/volumes: "true"
      spec:
        nodeSelector:
          cloud.google.com/gke-accelerator: "nvidia-l4"
        containers:
        - name: tfserve-server
          image: tensorflow/serving:2.13.1-gpu
  +       command: [ "tensorflow_model_server", "--model_name=$MODEL_NAME", "--model_base_path=/data/tfserve-model-repository/$MODEL_NAME", "--rest_api_port=8000", "--monitoring_config_file=$PATH_TO_MONITORING_CONFIG" ]
          ports:
          - name: http
            containerPort: 8000
          - name: grpc
            containerPort: 8500
          resources:
            limits:
              cpu: "4"
              memory: "16Gi"
              ephemeral-storage: "1Gi"
              nvidia.com/gpu: 1
            requests:
              cpu: "4"
              memory: "16Gi"
              ephemeral-storage: "1Gi"
              nvidia.com/gpu: 1
          volumeMounts:
          - name: gcs-fuse-csi-vol
            mountPath: /data
            readOnly: false
        serviceAccountName: $K8S_SA_NAME
        volumes:
        - name: gcs-fuse-csi-vol
          csi:
            driver: gcsfuse.csi.storage.gke.io
            readOnly: false
            volumeAttributes:
              bucketName: $GSBUCKET
              mountOptions: "implicit-dirs"
additional_install_info: |
  To verify that {{exporter_name}} is emitting metrics on the expected endpoints, do the following:

    1. Set up port forwarding by using the following command:
      <pre class="devsite-click-to-copy">
      kubectl -n {{namespace_name}} port-forward {{pod_name}} 8000
      </pre>
    2. Access the endpoint `localhost:8000/monitoring/prometheus/metrics` by using the browser
      or the `curl` utility in another terminal session.
podmonitoring_config: |
  apiVersion: monitoring.googleapis.com/v1
  kind: PodMonitoring
  metadata:
    name: tfserve
    labels:
      app.kubernetes.io/name: tfserve
      app.kubernetes.io/part-of: google-cloud-managed-prometheus
  spec:
    endpoints:
    - port: 8000
      scheme: http
      interval: 30s
      path: /monitoring/prometheus/metrics
    selector:
      matchLabels:
        app: tfserve
sample_promql_query: up{job="tfserve", cluster="{{cluster_name}}", namespace="{{namespace_name}}"}
dashboard_available: true
multiple_dashboards: false
dashboard_display_name: {{app_name}} Prometheus Overview