| 1 | plugin_name: go.d.plugin |
| 2 | modules: |
| 3 | - meta: |
| 4 | id: collector-go.d.plugin-nvidia_smi |
| 5 | plugin_name: go.d.plugin |
| 6 | module_name: nvidia_smi |
| 7 | monitored_instance: |
| 8 | name: Nvidia GPU |
| 9 | link: https://www.nvidia.com/en-us/ |
| 10 | icon_filename: nvidia.svg |
| 11 | categories: |
| 12 | - data-collection.hardware-and-sensors |
| 13 | keywords: |
| 14 | - nvidia |
| 15 | - gpu |
| 16 | - hardware |
| 17 | related_resources: |
| 18 | integrations: |
| 19 | list: |
| 20 | - plugin_name: go.d.plugin |
| 21 | module_name: dcgm |
| 22 | info_provided_to_referring_integrations: |
| 23 | description: "" |
| 24 | overview: |
| 25 | data_collection: |
| 26 | metrics_description: | |
| 27 | This collector monitors GPUs performance metrics using |
| 28 | the [nvidia-smi](https://developer.nvidia.com/nvidia-system-management-interface) CLI tool. |
| 29 | method_description: "" |
| 30 | supported_platforms: |
| 31 | include: [] |
| 32 | exclude: [] |
| 33 | multi_instance: true |
| 34 | additional_permissions: |
| 35 | description: "" |
| 36 | default_behavior: |
| 37 | auto_detection: |
| 38 | description: "" |
| 39 | limits: |
| 40 | description: "" |
| 41 | performance_impact: |
| 42 | description: "" |
| 43 | setup: |
| 44 | prerequisites: |
| 45 | list: [] |
| 46 | configuration: |
| 47 | file: |
| 48 | name: go.d/nvidia_smi.conf |
| 49 | options: |
| 50 | description: | |
| 51 | The following options can be defined globally: update_every, autodetection_retry. |
| 52 | folding: |
| 53 | title: Config options |
| 54 | enabled: true |
| 55 | list: |
| 56 | - name: update_every |
| 57 | description: Data collection frequency. |
| 58 | default_value: 10 |
| 59 | required: false |
| 60 | - name: autodetection_retry |
| 61 | description: Autodetection retry interval (seconds). Set 0 to disable. |
| 62 | default_value: 0 |
| 63 | required: false |
| 64 | - name: binary_path |
| 65 | description: Path to nvidia_smi binary. The default is "nvidia_smi" and the executable is looked for in the directories specified in the PATH environment variable. |
| 66 | default_value: nvidia_smi |
| 67 | required: false |
| 68 | - name: timeout |
| 69 | description: The maximum duration, in seconds, to wait for an `nvidia-smi` command to complete. This setting applies differently based on the collector's mode. **Loop Mode:** In loop mode, the timeout primarily determines how long to wait for the initial `nvidia-smi` execution. If the initial query takes longer than the timeout, the collector may report an error. For systems with multiple GPUs, the initial load time can sometimes be significant (e.g., 5-10 seconds). **Regular Mode:** If the collector is in regular mode, the timeout specifies how long to wait for each individual `nvidia-smi` execution. |
| 70 | default_value: 10 |
| 71 | required: false |
| 72 | - name: loop_mode |
| 73 | description: "When enabled, `nvidia-smi` is executed continuously in a separate thread using the `-l` option." |
| 74 | default_value: true |
| 75 | required: false |
| 76 | examples: |
| 77 | folding: |
| 78 | title: Config |
| 79 | enabled: true |
| 80 | list: |
| 81 | - name: Custom binary path |
| 82 | description: The executable is not in the directories specified in the PATH environment variable. |
| 83 | config: | |
| 84 | jobs: |
| 85 | - name: nvidia_smi |
| 86 | binary_path: /usr/local/sbin/nvidia_smi |
| 87 | troubleshooting: |
| 88 | problems: |
| 89 | list: [] |
| 90 | alerts: [] |
| 91 | metrics: |
| 92 | folding: |
| 93 | title: Metrics |
| 94 | enabled: false |
| 95 | description: "" |
| 96 | availability: [] |
| 97 | scopes: |
| 98 | - name: gpu |
| 99 | description: These metrics refer to the GPU. |
| 100 | labels: |
| 101 | - name: uuid |
| 102 | description: GPU uuid (e.g. GPU-27b94a00-ed54-5c24-b1fd-1054085de32a) |
| 103 | - name: index |
| 104 | description: GPU index (nvidia_smi typically orders GPUs by PCI bus ID) |
| 105 | - name: product_name |
| 106 | description: GPU product name (e.g. NVIDIA A100-SXM4-40GB) |
| 107 | metrics: |
| 108 | - name: nvidia_smi.gpu_pcie_bandwidth_usage |
| 109 | description: PCI Express Bandwidth Usage |
| 110 | unit: B/s |
| 111 | chart_type: line |
| 112 | dimensions: |
| 113 | - name: rx |
| 114 | - name: tx |
| 115 | - name: nvidia_smi.gpu_pcie_bandwidth_utilization |
| 116 | description: PCI Express Bandwidth Utilization |
| 117 | unit: '%' |
| 118 | chart_type: line |
| 119 | dimensions: |
| 120 | - name: rx |
| 121 | - name: tx |
| 122 | - name: nvidia_smi.gpu_fan_speed_perc |
| 123 | description: Fan speed |
| 124 | unit: '%' |
| 125 | chart_type: line |
| 126 | dimensions: |
| 127 | - name: fan_speed |
| 128 | - name: nvidia_smi.gpu_utilization |
| 129 | description: GPU utilization |
| 130 | unit: '%' |
| 131 | chart_type: line |
| 132 | dimensions: |
| 133 | - name: gpu |
| 134 | - name: nvidia_smi.gpu_memory_utilization |
| 135 | description: Memory utilization |
| 136 | unit: '%' |
| 137 | chart_type: line |
| 138 | dimensions: |
| 139 | - name: memory |
| 140 | - name: nvidia_smi.gpu_decoder_utilization |
| 141 | description: Decoder utilization |
| 142 | unit: '%' |
| 143 | chart_type: line |
| 144 | dimensions: |
| 145 | - name: decoder |
| 146 | - name: nvidia_smi.gpu_encoder_utilization |
| 147 | description: Encoder utilization |
| 148 | unit: '%' |
| 149 | chart_type: line |
| 150 | dimensions: |
| 151 | - name: encoder |
| 152 | - name: nvidia_smi.gpu_frame_buffer_memory_usage |
| 153 | description: Frame buffer memory usage |
| 154 | unit: B |
| 155 | chart_type: stacked |
| 156 | dimensions: |
| 157 | - name: free |
| 158 | - name: used |
| 159 | - name: reserved |
| 160 | - name: nvidia_smi.gpu_bar1_memory_usage |
| 161 | description: BAR1 memory usage |
| 162 | unit: B |
| 163 | chart_type: stacked |
| 164 | dimensions: |
| 165 | - name: free |
| 166 | - name: used |
| 167 | - name: nvidia_smi.gpu_temperature |
| 168 | description: Temperature |
| 169 | unit: Celsius |
| 170 | chart_type: line |
| 171 | dimensions: |
| 172 | - name: temperature |
| 173 | - name: nvidia_smi.gpu_voltage |
| 174 | description: Voltage |
| 175 | unit: V |
| 176 | chart_type: line |
| 177 | dimensions: |
| 178 | - name: voltage |
| 179 | - name: nvidia_smi.gpu_clock_freq |
| 180 | description: Clock current frequency |
| 181 | unit: MHz |
| 182 | chart_type: line |
| 183 | dimensions: |
| 184 | - name: graphics |
| 185 | - name: video |
| 186 | - name: sm |
| 187 | - name: mem |
| 188 | - name: nvidia_smi.gpu_power_draw |
| 189 | description: Power draw |
| 190 | unit: Watts |
| 191 | chart_type: line |
| 192 | dimensions: |
| 193 | - name: power_draw |
| 194 | - name: nvidia_smi.gpu_performance_state |
| 195 | description: Performance state |
| 196 | unit: state |
| 197 | chart_type: line |
| 198 | dimensions: |
| 199 | - name: P0-P15 |
| 200 | - name: nvidia_smi.gpu_mig_mode_current_status |
| 201 | description: MIG current mode |
| 202 | unit: status |
| 203 | chart_type: line |
| 204 | dimensions: |
| 205 | - name: enabled |
| 206 | - name: disabled |
| 207 | - name: nvidia_smi.gpu_mig_devices_count |
| 208 | description: MIG devices |
| 209 | unit: devices |
| 210 | chart_type: line |
| 211 | dimensions: |
| 212 | - name: mig |
| 213 | - name: mig |
| 214 | description: These metrics refer to the Multi-Instance GPU (MIG). |
| 215 | labels: |
| 216 | - name: uuid |
| 217 | description: GPU uuid (e.g. GPU-27b94a00-ed54-5c24-b1fd-1054085de32a) |
| 218 | - name: product_name |
| 219 | description: GPU product name (e.g. NVIDIA A100-SXM4-40GB) |
| 220 | - name: gpu_instance_id |
| 221 | description: GPU instance id (e.g. 1) |
| 222 | metrics: |
| 223 | - name: nvidia_smi.gpu_mig_frame_buffer_memory_usage |
| 224 | description: Frame buffer memory usage |
| 225 | unit: B |
| 226 | chart_type: stacked |
| 227 | dimensions: |
| 228 | - name: free |
| 229 | - name: used |
| 230 | - name: reserved |
| 231 | - name: nvidia_smi.gpu_mig_bar1_memory_usage |
| 232 | description: BAR1 memory usage |
| 233 | unit: B |
| 234 | chart_type: stacked |
| 235 | dimensions: |
| 236 | - name: free |
| 237 | - name: used |