Java

Agentless log submission is supported for the following languages and frameworks:

• dd-trace-java >= 1.35.2 and Log4j2.

Use the following environment variables to enable and configure Agentless log submission:

JavaScript/TypeScript

Agentless log submission is supported for the following tracer versions and logging libraries:

• dd-trace-js v4.48.0 or later on the v4 release line, or dd-trace-js v5.24.0 or later on the v5 release line, with winston.
• dd-trace-js v5.124.0 or later on the v5 release line, or dd-trace-js v6.13.0 or later on the v6 release line, with pino or bunyan.

Use the following environment variables to enable and configure Agentless log submission:

.NET

Agentless log submission is supported for the following languages and frameworks:

• dd-trace-dotnet >= 2.50.0 and XUnit TestOutputHelper.

Use the following environment variables to enable and configure Agentless log submission:

Swift

Use the following environment variables to enable and configure log submission:

Python

Requirements: ddtrace >= 4.8.0.

Log submission is supported for the pytest test framework, and only when logs are emitted with the standard library logging module.

Use the following environment variable to enable log submission for agentless mode:

If you use the Datadog Agent instead of agentless mode, set instead DD_LOGS_INJECTION=true in the environment.

Out-of-process logs

When a separate process executes code triggered by a test, it needs a trace_id and span_id from that test trace to correlate its logs. Use ddtrace.testing.logs.DDTestLogsHandler (ddtrace >= 4.11.0) to ship those log records to the Datadog logs intake, correlated with the originating test trace.

DDTestLogsHandler reads the same environment variables as the pytest plugin to detect the backend (agentless or EVP proxy). It is available in any subprocess where those variables are available.

Agentless mode (set DD_CIVISIBILITY_AGENTLESS_ENABLED=true):

Agent/EVP proxy mode (default):

Thread-per-worker

For one thread per test worker, use ThreadLocalCorrelationFilter to associate each thread’s log records with the correct test trace:

import logging
from ddtrace.testing.logs import DDTestLogsHandler, ThreadLocalCorrelationFilter

with DDTestLogsHandler(service="my-service") as handler:
    correlation = ThreadLocalCorrelationFilter()
    handler.addFilter(correlation)
    logging.getLogger().addHandler(handler)

    while True:
        job = queue.get()  # queue and run_test are provided by your worker framework
        correlation.set_context(trace_id=job.trace_id, span_id=job.span_id)
        run_test(job.item)

DDTestLogsHandler flushes buffered records automatically when used as a context manager. Call handler.close() if you do not use the context manager form.

Asyncio workers

For asyncio-based workers, ThreadLocalCorrelationFilter is not compatible with asyncio-based workers because thread-local storage does not propagate across asyncio.Task boundaries. Subclass CorrelationFilter and use a contextvars.ContextVar instead, which the event loop propagates automatically across await boundaries:

import asyncio
import contextvars
import logging
from ddtrace.testing.logs import CorrelationFilter, DDTestLogsHandler

class ContextVarCorrelationFilter(CorrelationFilter):
    def __init__(self):
        super().__init__()
        self._trace_id = contextvars.ContextVar("dd_trace_id", default=None)
        self._span_id = contextvars.ContextVar("dd_span_id", default=None)

    def set_context(self, trace_id, span_id):
        self._trace_id.set(trace_id)
        self._span_id.set(span_id)

    def get_trace_id(self):
        return self._trace_id.get()

    def get_span_id(self):
        return self._span_id.get()

async def run_one(job, correlation):
    correlation.set_context(trace_id=job.trace_id, span_id=job.span_id)
    await run_test(job.item)

async def main(jobs):
    with DDTestLogsHandler(service="my-service") as handler:
        correlation = ContextVarCorrelationFilter()
        handler.addFilter(correlation)
        logging.getLogger().addHandler(handler)
        await asyncio.gather(*(run_one(job, correlation) for job in jobs))

Ruby

Agentless logs submission with Test Optimization is supported for Rails applications. Before enabling, ensure that your application is instrumented with Datadog tracing.

To use agentless log submission, you need datadog-ci version 0.16 or later. The following logging libraries are supported:

• activesupport >= 5.0 (only when using ActiveSupport::TaggedLogging)
• lograge >= 0.14
• semantic_logger >= 4.0

Use the following environment variable to enable log submission:

1. Set up log collection through the Datadog Agent.
2. Follow the steps described in Correlate Logs and Traces.