Datadog CI CLI

To enable direct links from Datadog errors to your source code (e.g., GitHub, GitLab, Bitbucket, Azure DevOps), the upload command must be able to associate sourcemaps with your repository.

Requirements for automatic linking:

1. The git executable must be installed.
2. datadog-ci must be run within the git repository.
3. --disable-git must not be set.

Manual Configuration (Bypassing Git): If your build environment lacks a .git/ folder, you can still enable linking by providing:

• --repository-url <url> (or DD_GIT_REPOSITORY_URL)
• --commit-sha <sha> (or DD_GIT_COMMIT_SHA)

Note: In this mode, source files are extracted directly from the sourcemap sources field, so untracked files may be sent to Datadog.

Handling Path Prefixes: If your sourcemap references paths with a prefix before the repository root (e.g., webpack://MyProject/src/file.js), use the --project-path MyProject/ flag so files are correctly matched to the repository.

The --ignored-source-paths option (or the DD_COVERAGE_IGNORED_SOURCE_PATHS environment variable) excludes specific source files from the coverage computation.

Important Behaviors

• Replacement, not Merging: This option replaces the ignore list defined in code-coverage.datadog.yml. If this option is not provided, the configuration file's list is used.
• Server-side Matching: Patterns are sent verbatim to Datadog and matched server-side; they are not expanded against your local filesystem.
• Pattern Syntax:
  • Patterns are separated by commas or newlines.
  • Use brace groups {} to protect commas from being treated as separators (e.g., **/*.{js,ts}).
  • A comma inside a character class (e.g., [a,b]) is still treated as a separator; use [ab] instead.
• Limits: Up to 2,000 patterns and 256 KB total are accepted. If you exceed 1,000 patterns or 100 KB, the upload proceeds but prints a warning.

Platform-specific Argument Limits

If your list is very large, be aware of OS limits:

• Linux: 128 KB limit for arguments/environment variables.
• Windows: ~8,191 character limit via cmd.exe (used by npm/yarn shims).
• macOS: No per-argument limit, but shares a budget with the environment (256 KB limit applies).

# Exclude generated files using glob patterns
datadog-ci coverage upload --ignored-source-paths "**/generated/**,src/gen/**" .

The --tracing flag controls Application Performance Monitoring (APM) instrumentation.

Available values:

• inject: Automatically detects the application language and injects the appropriate tracer.
• manual: Use this if your application image already includes a tracer (e.g., for Go applications using dd-trace-go).
• disabled: Turns tracing off.
• true/1: Alias for manual.
• false/0: Alias for disabled.

Note: If you omit the --tracing flag, the command preserves existing automatic instrumentation and DD_TRACE_ENABLED values, or enables tracing if no value currently exists.

# Enable automatic APM instrumentation with language detection
datadog-ci cloud-run instrument -p <gcp-project> -r us-central1 -s <service-name> --tracing inject

# Use manual mode if the tracer is already in the image
datadog-ci cloud-run instrument -p <gcp-project> -r us-central1 -s <service-name> --tracing manual

You can automatically inject language-specific tracers into your Container App using the --tracing inject flag. This process uses an init container to copy the tracer into your application without rebuilding the image.

Key Behaviors:

• Language Detection: Using --tracing inject without a --language flag will attempt to detect the application language automatically. This uses a larger composite tracer image.
• Single Language Injection: To reduce image size and potential cold-start delays (especially for scale-to-zero workloads), specify a single language using --language <language>. Supported values are: java, nodejs, csharp, python, ruby, and php.
• Go Support: Go tracers cannot be injected automatically. You must install dd-trace-go in your application image and use --tracing manual.
• Manual Tracing: If the tracer is already installed in your application image, use --tracing manual.
• Container Selection: If your Container App has multiple application containers, use --container-name to target a specific one. Note that --name refers to the Container App resource itself.

Constraints:

• Ruby injection does not support musl.
• .NET tracer versions before 3.0 are not supported.
• --tracer-version and --tracer-libc require the --language flag.

# Detect the application language and add its tracer automatically
datadog-ci container-app instrument \
  --subscription-id <subscription-id> \
  --resource-group <resource-group-name> \
  --name <container-app-name> \
  --tracing inject

# Add only the Python tracer
datadog-ci container-app instrument \
  --subscription-id <subscription-id> \
  --resource-group <resource-group-name> \
  --name <container-app-name> \
  --tracing inject \
  --language python

You can enrich Datadog errors with direct links to your source code in GitHub, GitLab, Bitbucket, or Azure DevOps.

Requirements

• The git executable must be installed.
• datadog-ci must be run from within the git repository.

When these are met, the command automatically gathers the commit hash, repository URL, and tracked file paths.

Configuration

• Repository URL: Inferred from the origin remote. Use --repository-url to override this (e.g., if using a custom GitLab URL).
• Project Path: If running the command outside the React Native project root, use --project-path to provide the absolute path to the project root. This ensures absolute paths in the source maps are correctly mapped to your local files.

Supported Providers

Datadog generates links for:

• GitHub/GitLab: https://<url>/blob/<commit>/<path>#L<line>
• Bitbucket: https://<url>/src/<commit>/<path>#lines-<line>
• Azure DevOps: Uses a specific query parameter format for version and path.