For AI agents: The documentation index is at https://docs.digitalocean.com/llms.txt. Markdown versions of pages use the same URL with index.html.md in place of the HTML page (for example, append index.html.md to the directory path instead of opening the HTML document).

Metrics Agent

The metrics agent runs as an unprivileged user. It reads system metrics from /proc, stores authentication data in /var/opt, and stores the agent binary in /opt/digitalocean.

We support Ubuntu 14.04 or later, CentOS 6 or later, Debian 8 or later, and Fedora 27 or later. After you install the agent, Monitoring shows enhanced graphs, supports resource alerts, and sends metrics to the Monitoring service so you can track resource usage and troubleshoot performance issues.

To help improve Monitoring, the metrics agent reports process names, but it does not collect environment variables or process arguments. You can also opt out of process name collection.

The metrics agent uses outbound connections only. It uses port 80 to contact the DigitalOcean metadata service for an authentication token and port 443 to send encrypted metrics data. It does not open any ports or interfere with existing services.

Droplet Observability

Droplet observability shows system-level metrics over time. Use them to monitor resource usage and understand how it relates to Droplet performance.

By default, Droplet metrics show CPU usage, public and private bandwidth usage, and disk I/O. If you install the DigitalOcean metrics agent, you also get load averages, memory usage, and disk usage.

GPU Observability

GPU Observability extends DigitalOcean Insights with GPU-level metrics for DOKS clusters that use GPU node pools created from AI/ML Ready images for AMD or NVIDIA GPUs. It lets you monitor GPU workloads directly from the Insights tab, including utilization, temperature, memory usage, and performance metrics. AMD GPU support is available by request only.

do-agent automatically detects the GPU type on each node and enables the appropriate exporter: DCGM for NVIDIA GPUs and ROCm for AMD GPUs. It collects metrics locally on each GPU worker node.

GPU Observability is available on DOKS version 1.33.1-do.5 and later. It is enabled automatically when you select Improved metrics and monitoring during cluster creation.

For security, the GPU exporters listen only on 127.0.0.1, which prevents external access.

Resource Alerts

Resource alerts notify you when Droplet metrics cross thresholds you define, and you can receive notifications by email, Slack, or both. Alerts support CPU utilization, load average, memory utilization, disk utilization, disk I/O, and public or private bandwidth.

When an alert triggers, it appears in the DigitalOcean Control Panel under the Resource Alerts page with the alert name, the affected Droplet or Droplets, and links to their pages. You also receive a notification with the alert details and usage information, followed by another notification when the condition clears.

Resource alerts use data from the metrics agent. To use resource alerts, you must install the metrics agent on the Droplets or tags you want to monitor. Kubernetes worker nodes include the metrics agent by default.