Nvidia SMI and Zabbix Integration

Input and output integration overview

The Nvidia SMI Plugin enables the retrieval of detailed statistics about NVIDIA GPUs attached to the host system, providing essential insights for performance monitoring.

This plugin sends metrics to Zabbix via traps, allowing for efficient monitoring of systems and applications. It supports automated configuration and data sending based on dynamic metrics collected by Telegraf.

Integration details

Nvidia SMI

The Nvidia SMI Plugin is designed to gather metrics regarding the performance and status of NVIDIA GPUs on the host machine. By leveraging the capabilities of the nvidia-smi command-line tool, this plugin pulls crucial information such as GPU memory utilization, temperature, fan speed, and various performance metrics. This data is essential for monitoring GPU health and performance in real-time, particularly in environments where GPU performance directly impacts computing tasks, such as machine learning, 3D rendering, and high-performance computing. The plugin provides flexibility by allowing users to specify the path to the nvidia-smi binary and configure polling timeouts, accommodating both Linux and Windows systems where the nvidia-smi tool is commonly located. With its ability to collect detailed statistics on each GPU, this plugin becomes a vital resource for any infrastructure relying on NVIDIA hardware, facilitating proactive management and performance tuning.

Zabbix

The Telegraf Zabbix plugin is designed to send metrics to Zabbix, an open-source monitoring solution, using the trap protocol. It supports various versions from 3.0 to 6.0, ensuring compatibility with recent updates. The plugin facilitates easy integration with the Zabbix ecosystem, allowing users to send collected metrics and monitor system performance seamlessly. Key functionalities include the ability to define the address and port of the Zabbix server, options for prefixing keys, determining the type of data sent (active vs. trapper), and features for low-level discovery (LLD) enabling dynamic item creation based on the metrics observed. Configuration options also allow for autoregistration and resending intervals for LLD data, ensuring that the metrics are up-to-date and relevant. Additionally, the trap format used for sending metrics is structured to facilitate efficient data transfer and processing in Zabbix.

Configuration

Nvidia SMI

[[inputs.nvidia_smi]]
  ## Optional: path to nvidia-smi binary, defaults "/usr/bin/nvidia-smi"
  ## We will first try to locate the nvidia-smi binary with the explicitly specified value (or default value),
  ## if it is not found, we will try to locate it on PATH(exec.LookPath), if it is still not found, an error will be returned
  # bin_path = "/usr/bin/nvidia-smi"

  ## Optional: timeout for GPU polling
  # timeout = "5s"

Zabbix

[[outputs.zabbix]]
  ## Address and (optional) port of the Zabbix server
  address = "zabbix.example.com:10051"

  ## Send metrics as type "Zabbix agent (active)"
  # agent_active = false

  ## Add prefix to all keys sent to Zabbix.
  # key_prefix = "telegraf."

  ## Name of the tag that contains the host name. Used to set the host in Zabbix.
  ## If the tag is not found, use the hostname of the system running Telegraf.
  # host_tag = "host"

  ## Skip measurement prefix to all keys sent to Zabbix.
  # skip_measurement_prefix = false

  ## This field will be sent as HostMetadata to Zabbix Server to autoregister the host.
  ## To enable this feature, this option must be set to a value other than "".
  # autoregister = ""

  ## Interval to resend auto-registration data to Zabbix.
  ## Only applies if autoregister feature is enabled.
  ## This value is a lower limit, the actual resend should be triggered by the next flush interval.
  # autoregister_resend_interval = "30m"

  ## Interval to send LLD data to Zabbix.
  ## This value is a lower limit, the actual resend should be triggered by the next flush interval.
  # lld_send_interval = "10m"

  ## Interval to delete stored LLD known data and start capturing it again.
  ## This value is a lower limit, the actual resend should be triggered by the next flush interval.
  # lld_clear_interval = "1h"

Input and output integration examples

Nvidia SMI

1. Real-Time GPU Monitoring for ML Training: Continuously monitor the GPU utilization and memory usage during machine learning model training. This enables data scientists to ensure that their GPUs are not being overutilized or underutilized, optimizing resource allocation and reviewing performance bottlenecks in real-time.

2. Automated Alerts for Overheating GPUs: Implement a system using the Nvidia SMI plugin to track GPU temperatures and set alerts for instances where temperatures exceed safe thresholds. This proactive monitoring can prevent hardware damage and improve system reliability by alerting administrators to potential cooling issues before they result in failure.

3. Performance Baselines for GPU Resources: Establish baseline performance metrics for your GPU resources. By regularly collecting data and analyzing trends in GPU usage, organizations can identify anomalies and optimize their workloads accordingly, leading to enhanced operational efficiency.

4. Dockerized GPU Usage Insights: In a containerized environment, use the plugin to monitor GPU performance from within a Docker container. This allows developers to track GPU performance of their applications in production, facilitating troubleshooting and performance optimization within isolated environments.

Zabbix

1. Dynamic Monitoring of Containerized Applications: Integration of the Zabbix plugin can be leveraged to monitor Docker containers dynamically. As containers are created and removed, the plugin can automatically update Zabbix with the appropriate metrics, ensuring that monitoring stays current without manual configuration. This enhances visibility into resource usage and performance metrics for microservices orchestrated with Kubernetes or Docker Swarm.

2. Real-Time Performance Monitoring with Auto-registration: By enabling the autoregister feature, the plugin can automatically register hosts in Zabbix based on the metrics received. This scenario provides a streamlined approach to add new hosts to monitoring without manual setup, which is particularly useful in environments where hosts may frequently spin up and down, such as serverless architectures or cloud-based deployments.

3. Leveraging Low-level Discovery for Flexible Metric Capture: Using low-level discovery, this plugin allows Zabbix to adaptively create items for metrics that are not predefined. In a scenario involving multiple network devices reporting different performance metrics, the plugin can dynamically inform Zabbix about new metrics as they appear, thus ensuring comprehensive monitoring capabilities that evolve with the monitored systems.

4. Centralized Monitoring of Distributed Systems: The Zabbix plugin can be utilized in a centralized monitoring setup for distributed systems where multiple Telegraf instances are running across different geographical locations. By sending all metrics to a central Zabbix server, organizations can achieve a holistic view of their infrastructure’s performance and make informed operational decisions.

Feedback

Thank you for being part of our community! If you have any general feedback or found any bugs on these pages, we welcome and encourage your input. Please submit your feedback in the InfluxDB community Slack.