doca-telemetry-utils

Discover DOCA diagnostic counters, resolve Data IDs, and validate per-device telemetry support.

3.2k|370|Updated Feb 25, 2026
One-click install
npx skills add https://github.com/NVIDIA/skills --skill doca-telemetry-utils
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: doca-telemetry-utils
Source: https://github.com/NVIDIA/skills/tree/main/skills/doca-telemetry-utils
Command: npx skills add https://github.com/NVIDIA/skills --skill doca-telemetry-utils

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Operators standing up a DOCA Telemetry exporter/collector pipeline often hit silent metric drops because exporter configs reference counter names instead of the binary Data IDs the exporter actually ships, or because a counter is not supported on the target BlueField device. This Skill guides agents through the doca_telemetry_utils CLI to discover the counter schema, translate names to Data IDs, and validate per-device support before committing an exporter config.

Core Features & Use Cases

  • Schema discovery: Enumerate every diagnostic counter the installed DOCA Telemetry surface knows about via doca_telemetry_utils get-counters.
  • Name ↔ Data ID translation: Resolve counter names plus property dimensions (node, pcie_index, depth) to binary Data IDs, and reverse-resolve captured Data IDs back to names and properties.
  • Per-device validation: Probe a specific BlueField PCI address to confirm a counter is supported before committing it to an exporter config, preventing silent metric drops.
  • Use Case: An operator reports "my exporter ships but the collector sees nothing." The agent walks the layered error taxonomy, re-resolves the counter, probes the device, and identifies the unsupported counter causing the silent drop.

Quick Start

Ask your agent to enumerate the DOCA diagnostic counters on this host and validate that a chosen counter is supported on your BlueField before adding it to the exporter config.

Frequently Asked Questions about doca-telemetry-utils

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I translate a DOCA telemetry counter name to a Data ID?

Run doca_telemetry_utils with the counter name and property values to resolve the Data ID, name, unit, and property tuple. First run the tool with only the counter name to print the documented property options, then re-run with chosen values.

Why does my DOCA exporter ship but the collector sees nothing?

The most common cause is a counter name in the exporter config that resolves to a Data ID the target device does not support, so the metric is silently dropped. Resolve each counter through doca_telemetry_utils and run the per-device probe before committing the config.

How do I check which counters a BlueField device supports?

Run doca_telemetry_utils get-counters to enumerate the schema DOCA knows about, then pass the device PCI address with the counter name to probe per-device support. Schema knowledge and per-device support are separate checks.

Can I reverse-resolve a captured Data ID back to a counter name?

Yes, pass the 0x-prefixed hexadecimal Data ID to doca_telemetry_utils to print the counter name, unit, and encoded property values. Data IDs must come from a forward resolve on the same DOCA version, not be hand-constructed.

Does doca_telemetry_utils configure the telemetry exporter or DTS?

No, it is a read-only operator-side support tool for schema discovery and per-device validation. Exporter/collector programming belongs to the doca-telemetry library skill, and DTS deployment is routed through the public DOCA documentation.

Do I need to re-resolve counters after a DOCA upgrade?

Yes, the counter catalog is install-versioned and names or property axes can change across DOCA releases. Re-run get-counters and re-resolve every counter in exporter configs against the new install before rolling the pipeline forward.