integration_test/third_party_apps_test/applications/dcgmv1/metadata.yaml

# Copyright 2022 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.

app_url: "https://developer.nvidia.com/dcgm"
short_name: NVIDIA DCGM
long_name: NVIDIA Data Center GPU Manager (DCGM)
logo_path: /stackdriver/images/integrations/nvidia.png # supplied by google technical writer
description: |-
  The NVIDIA Data Center GPU Manager (DCGM) integration collects advanced GPU metrics,
  including SM block utilization, Pipe utilization, PCIe and NVLink traffic.
configure_integration: |-
  You must install DCGM and run the DCGM daemon service.
supported_operating_systems: linux
supported_app_version: ["3.1"]
gpu_platforms: # p4, p100 don't emit DCGM profiling metrics
  - model: a100
    platforms:
      - ubuntu-os-cloud:ubuntu-2004-lts
  - model: v100
    platforms:
      - ubuntu-os-cloud:ubuntu-2004-lts
  - model: t4
    platforms:
      - ubuntu-os-cloud:ubuntu-2004-lts
  - model: l4
    platforms:
      - debian-cloud:debian-11
      - ml-images:common-gpu-debian-11-py310
      - rocky-linux-cloud:rocky-linux-8
      - rocky-linux-cloud:rocky-linux-9
      - suse-cloud:sles-15
      - ubuntu-os-cloud:ubuntu-2004-lts
      - ubuntu-os-cloud:ubuntu-2204-lts
  - model: h100
    platforms:
      - ubuntu-os-cloud:ubuntu-minimal-2004-lts # due to H100 quota, choose an image from the exhaustive list to skip presubmits
expected_metrics:
  - type: workload.googleapis.com/dcgm.gpu.profiling.sm_utilization
    value_type: DOUBLE
    kind: GAUGE
    description: Fraction of time at least one warp was active on a multiprocessor, averaged over all multiprocessors.
    monitored_resources: [gce_instance]
    labels:
      - &model_label
        name: model
        value_regex: .*
        description: GPU model name.
      - &uuid_label
        name: uuid
        value_regex: .*
        description: GPU universally unique identifier.
      - &gpu_number_label
        name: gpu_number
        value_regex: "[0-9]*"
        description: GPU index starting at 0.
      - &instrumentation_source_label
        name: instrumentation_source
        value_regex: agent.googleapis.com/dcgm
    notes:
      - &dcgm_profiling_note Not available on GPU models P100 and P4.
    representative: true
  - type: workload.googleapis.com/dcgm.gpu.profiling.sm_occupancy
    value_type: DOUBLE
    kind: GAUGE
    description: Fraction of the number of warps resident on a multiprocessor, averaged over all multiprocessors.
    monitored_resources: [gce_instance]
    labels:
      - *model_label
      - *uuid_label
      - *gpu_number_label
      - *instrumentation_source_label
    notes:
      - *dcgm_profiling_note
  - type: workload.googleapis.com/dcgm.gpu.profiling.pipe_utilization
    value_type: DOUBLE
    kind: GAUGE
    description: Fraction of cycles the corresponding GPU pipe was active, averaged over time and all multiprocessors.
    monitored_resources: [gce_instance]
    labels:
      - *model_label
      - *uuid_label
      - *gpu_number_label
      - name: pipe
        value_regex: tensor|fp64|fp32|fp16
        description: "GPU pipe in use, one of [tensor, fp64, fp32, fp16]."
        notes:
          - For L4, the `pipe` value `fp64` is not supported.
      - *instrumentation_source_label
    notes:
      - *dcgm_profiling_note
  - type: workload.googleapis.com/dcgm.gpu.profiling.dram_utilization
    value_type: DOUBLE
    kind: GAUGE
    description: Fraction of cycles data was being sent or received from GPU memory.
    monitored_resources: [gce_instance]
    labels:
      - *model_label
      - *uuid_label
      - *gpu_number_label
      - *instrumentation_source_label
  - type: workload.googleapis.com/dcgm.gpu.profiling.pcie_traffic_rate
    value_type: INT64
    kind: GAUGE
    description: The number of bytes sent over the PCIe bus, including both protocol headers and data payloads.
    unit: By/s
    monitored_resources: [gce_instance]
    labels:
      - *model_label
      - *uuid_label
      - *gpu_number_label
      - &direction_label
        name: direction
        value_regex: tx|rx
        description: "Direction of the link traffic, one of [tx, rx]."
      - *instrumentation_source_label
    notes:
      - *dcgm_profiling_note
  - type: workload.googleapis.com/dcgm.gpu.profiling.nvlink_traffic_rate
    value_type: INT64
    kind: GAUGE
    description: The number of bytes sent over NVLink, not including protocol headers.
    unit: By/s
    monitored_resources: [gce_instance]
    labels:
      - *model_label
      - *uuid_label
      - *gpu_number_label
      - *direction_label
      - *instrumentation_source_label
    notes:
      - *dcgm_profiling_note
configuration_options:
  metrics:
    - type: dcgm
      fields:
        - name: type
          default: null
          description: This value must be `dcgm`.
        - name: collection_interval
          default: 60s
          description: A [time duration](https://pkg.go.dev/time#ParseDuration) value, such as `30s` or `5m`.
        - name: receiver_version
          default: "1"
          description: Either 1 or 2. Version 2 has many more metrics available.
        - name: endpoint
          default: localhost:5555
          description: The DCGM service endpoint specified as `hostname:port`.
minimum_supported_agent_version:
  metrics: 2.38.0
public_url: https://cloud.google.com/stackdriver/docs/solutions/agents/ops-agent/third-party-nvidia