nv-one-logger-core

What it is and what it does

nv-one-logger-core is a telemetry collection library that provides the foundational abstractions for instrumenting applications—particularly long-running jobs like ML training. It defines core concepts (spans representing units of work, events representing singular points in time, and attributes as properties) modeled on OpenTelemetry standards. The library intentionally keeps its core lightweight by minimizing third-party dependencies, allowing it to be adopted without conflicting with existing application dependencies.

The package provides a Recorder interface that applications use to start and stop spans, record events, and report errors. It also includes a flexible exporter configuration system supporting direct code configuration, YAML/JSON files, and Python entry points, with clear priority rules. Applications can use it via higher-level domain-specific libraries, the Recorder interface directly, a context manager (timed_span), or by instantiating exporters manually—though the Recorder approach is recommended to reduce instrumentation mistakes.

Use it for:

• Instrument ML training jobs to track checkpointing, model evaluation, and other domain-specific events with automatic timing
• Collect distributed traces from microservices by creating spans for operations and exporting them to a collector pipeline
• Add telemetry to long-running batch jobs with configurable exporters (Kafka, Weights & Biases, OTEL collector) without tight coupling
• Build domain-specific observability libraries on top of the core abstractions for specialized application types
• Aggregate and filter telemetry events at the recorder level (e.g., computing checkpoint statistics across multiple events)

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides core telemetry abstractions (spans, events, attributes) and a recorder interface for collecting and exporting observability data from applications and long-running jobs.

Yes, if you need a lightweight, vendor-agnostic telemetry foundation for applications or jobs. The core is intentionally minimal to avoid dependency conflicts. However, the aging maintenance status (289 days since last release) suggests limited active development—verify whether it fits your stability and support expectations before committing to it as a long-term instrumentation backbone.

Install

nv-one-logger-core on PyPI

pip install nv-one-logger-core

uv add nv-one-logger-core

poetry add nv-one-logger-core

Installing nv-one-logger-core

Before you install

Low install friction with a pure-Python wheel distribution. Maintenance status is aging (289 days since last release), which may indicate the package is stable but not actively developed.

License in practice

Licensed under Apache-2.0 (permissive), allowing use in most commercial and open-source projects with minimal restrictions.

Quickstart

pip install nv-one-logger-core

from nv_one_logger.api.timed_span import configure_one_logger, get_recorder, timed_span
from nv_one_logger.core import Attributes

with timed_span(name="operation", span_attributes=Attributes({"tag": "value"})):
    # business logic here
    pass

Requires Python 3.8 or later (supports up to 3.13). Runtime dependencies (pydantic, overrides, StrEnum, toml, typing-extensions) must be available.

Verify before relying

• Whether the package is actively maintained or in stable/maintenance-only mode despite the aging status
• Real-world performance characteristics when handling high-volume telemetry from large-scale jobs

Package facts

Evidence: nv_one_logger_core-2.3.1-py3-none-any.whl

Tags

More Monitoring packages