Datadog APM test agent

The APM test agent is an application which emulates the APM endpoints of the Datadog agent which can be used for testing Datadog APM client libraries.

See the Features section for the complete list of functionalities provided.

See the HTTP API section for the endpoints available.

See the Development section for how to get the test agent running locally to add additional checks or fix bugs.

Installation

The test agent can be installed from PyPI:

pip install ddapm-test-agent

ddapm-test-agent --port=8126

or from Docker:

# Run the test agent and mount the snapshot directory
docker run --rm\
        -p 8126:8126\
        -e SNAPSHOT_CI=0\
        -v $PWD/tests/snapshots:/snapshots\
        ghcr.io/datadog/dd-apm-test-agent/ddapm-test-agent:latest

or from source:

pip install git+https://github.com/Datadog/dd-apm-test-agent

or a specific branch:

pip install git+https://github.com/Datadog/dd-apm-test-agent@{branch}

Features

Trace invariant checks

Many checks are provided by the test agent which will verify trace data. All checks are enabled by default and can be manually disabled.

See the configuration section for the options.

Check description	Check name
Trace count header matches number of traces	`trace_count_header`
Client library version header included in request	`meta_tracer_version_header`
Trace content length header matches payload size	`trace_content_length`

Returning data

All data that is submitted to the test agent can be retrieved.

Traces can be returned via the /test/traces endpoint documented below.

Helpful logging

The INFO log level of the test agent outputs useful information about the requests the test agent receives. For traces this includes a visual representation of the traces.

INFO:ddapm_test_agent.agent:received trace payload with 1 trace chunk
INFO:ddapm_test_agent.agent:Chunk 0
[parent]
├─ [child1]
├─ [child2]
└─ [child3]
INFO:ddapm_test_agent.agent:end of payload ----------------------------------------

Proxy

The test agent provides proxying to the Datadog agent. This is enabled by passing the agent url to the test agent either via the --agent-url command-line argument or by the DD_TRACE_AGENT_URL or DD_AGENT_URL environment variables.

When proxying is enabled, the response from the Datadog agent will be returned instead of one from the test agent.

At the trace-level, proxying can also be disabled by including the X-Datadog-Agent-Proxy-Disabled header with a value of true. This will disable proxying after a trace is handled, regardless of whether an agent URL is set.

Snapshot testing

The test agent provides a form of characterization testing which we refer to as snapshotting. This allows library maintainers to ensure that traces don't change unexpectedly when making unrelated changes.

This can be used to write integration tests by having test cases use the tracer to emit traces which are collected by the test agent and compared against reference traces stored previously.

To do snapshot testing with the test agent:

Ensure traces are associated with a session token (typically the name of the test case) by either:
- Calling the /test/session/start with the token endpoint before emitting the traces; or
- Attaching an additional query string parameter or header specifying the session token on /vX.Y/trace requests (see below for the API specifics). (Required for concurrent test running)
Emit traces (run the integration test).
Signal the end of the session and perform the snapshot comparison by calling the /tests/session/snapshot endpoint with the session token. The endpoint will return a 400 response code if the snapshot failed along with a plain-text trace of the error which can be forwarded to the test framework to help triage the issue.

Snapshot output

The traces are normalized and output in JSON to a file. The following transformations are made to the input:

Trace ids are overwritten to match the order in which the traces were received.
Span ids are overwritten to be the DFS order of the spans in the trace tree.
Parent ids are overwritten using the normalized span ids. However, if the parent is not a span in the trace, the parent id is not overwritten. This is necessary for handling distributed traces where all spans are not sent to the same agent.
Span attributes are ordered to be more human-readable, with the important attributes being listed first.
Span attributes are otherwise ordered alphanumerically.
The span meta and metrics maps if empty are excluded.

Configuration

The test agent can be configured via command-line options or via environment variables.

Command line

ddapm-test-agent

ddapm-test-agent is command used to run a test agent.

Please refer to ddapm-test-agent --help for more information.

ddapm-test-agent-fmt

ddapm-test-agent-fmt is a command line tool to format or lint snapshot json files.

# Format all snapshot json files
ddapm-test-agent-fmt path/to/snapshots

# Lint snapshot json files
ddapm-test-agent-fmt --check path/to/snapshots

Please refer to ddapm-test-agent-fmt --help for more information.

Environment Variables

PORT [8126]: Port to listen on.
ENABLED_CHECKS [""]: Comma-separated values of checks to enable. Valid values can be found in trace invariant checks
LOG_LEVEL ["INFO"]: Log level to use. DEBUG, INFO, WARNING, ERROR, CRITICAL.
LOG_SPAN_FMT ["[{name}]"]: Format string to use when outputting spans in logs.
SNAPSHOT_DIR ["./snapshots"]: Directory in which snapshots will be stored. Can be overridden by providing the dir query parameter on /snapshot.
SNAPSHOT_CI [0]: Toggles CI mode for the snapshot tests. Set to 1 to enable. CI mode does the following:
- When snapshots are unexpectedly generated from a test case a failure will be raised.
SNAPSHOT_IGNORED_ATTRS ["span_id,trace_id,parent_id,duration,start,metrics.system.pid,metrics.process_id,metrics.system.process_id,meta.runtime-id"]: The attributes to ignore when comparing spans in snapshots.
DD_AGENT_URL [""]: URL to a Datadog agent. When provided requests will be proxied to the agent.
DD_APM_RECEIVER_SOCKET [""]: When provided, the test agent will listen for traces on a socket at the path provided (e.g., /var/run/datadog/apm.socket)
DD_SUPPRESS_TRACE_PARSE_ERRORS [false]: Set to "true" to disable span parse errors when decoding handled traces. When disabled, errors will not be thrown for metrics incorrectly placed within the meta field, or other type errors related to span tag formatting/types. Can also be set using the --suppress-trace-parse-errors=true option.
SNAPSHOT_REMOVED_ATTRS [""]: The attributes to remove from spans in snapshots. This is useful for removing attributes that are not relevant to the test case. Note that removing span_id is not permitted to allow span ordering to be maintained.
DD_POOL_TRACE_CHECK_FAILURES [false]: Set to "true" to pool Trace Check failures that occured within Test-Agent memory. These failures can be queried later using the /test/trace_check/failures endpoint. Can also be set using the --pool-trace-check-failures=true option.
DD_DISABLE_ERROR_RESPONSES [false]: Set to "true" to disable Test-Agent <Response 400> when a Trace Check fails, instead sending a valid <Response 200>. Recommended for use with the DD_POOL_TRACE_CHECK_FAILURES env variable. Can also be set using the --disable-error-responses=true option.

HTTP API

/test/traces

Return traces that have been received by the agent. Traces matching specific trace ids can be requested with the options below.

[optional] `?trace_ids=`

[optional] `X-Datadog-Trace-Ids`

Specify trace ids as comma separated values (eg. 12345,7890,2468)

/test/session/start

Initiate a synchronous session. All subsequent traces received will be associated with the required test token provided.

[optional] `?agent_sample_rate_by_service=`

Sample rates to be returned by the agent in response to trace v0.4 and v0.5 requests.

Example: "{'service:test,env:staging': 0.5, 'service:test2,env:prod': 0.2}" (note the JSON has to be URL-encoded).

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

Test session token for a test case. Ensure this value is unique to avoid conflicts between sessions.

/test/session/snapshot

Perform a snapshot generation or comparison on the data received during the session.

Snapshots are generated when the test agent is not in CI mode and there is no snapshot file present. Otherwise a snapshot comparison will be performed.

[optional*] `?test_session_token=`

[optional*] `X-Datadog-Test-Session-Token`

To run test cases in parallel this HTTP header must be specified. All test cases sharing a test token will be grouped.

* Required for concurrent tests. Either via query param or HTTP header.

[optional] `?ignores=`

Comma-separated list of keys of which to ignore values for.

The default built-in ignore list is: span_id, trace_id, parent_id, duration, start, metrics.system.pid, metrics.process_id, metrics.system.process_id, meta.runtime-id.

[optional] `?dir=`

default: ./snapshots (relative to where the test agent is run).

Override the directory where the snapshot will be stored and retrieved from. This directory must already exist.

This value will override the environment variable SNAPSHOT_DIR.

Warning: it is an error to specify both dir and file.

[optional] `?file=`

[optional] `X-Datadog-Test-Snapshot-Filename`

An absolute or relative (to the current working directory of the agent) file name where the snap will be stored and retrieved.

Warning: it is an error to specify both file and dir.

Note: the file extension will be appended to the filename.

_tracestats will be appended to the filename for trace stats requests.

[optional] `?removes=`

Comma-separated list of keys that will be removed from spans in the snapshot.

The default built-in remove list does not remove any keys.

/test/session/requests

Return all requests that have been received by the agent for the given session token.

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

Returns the requests in the following json format:

[
  {
    "headers": {},
    "body": "...",
    "url": "http...",
    "method": "GET"
  }
]

body is a base64 encoded body of the request.

/test/session/traces

Return traces that have been received by the agent for the given session token.

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

/test/session/stats

Return stats that have been received by the agent for the given session token.

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

Stats are returned as a JSON list of the stats payloads received.

/test/session/responses/config (POST)

Create a Remote Config payload to retrieve in endpoint /v0.7/config

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

curl -X POST 'http://0.0.0.0:8126/test/session/responses/config/path' -d '{"roots": ["eyJ....fX0="], "targets": "ey...19", "target_files": [{"path": "datadog/2/ASM_DATA/blocked_users/config", "raw": "eyJydWxlc19kYXRhIjogW119"}], "client_configs": ["datadog/2/ASM_DATA/blocked_users/config"]}'

/test/session/responses/config/path (POST)

Due to Remote Config payload being quite complicated, this endpoint works like /test/session/responses/config (POST) but you should send a path and a message and this endpoint builds the Remote Config payload.

The keys of the JSON body are path and msg

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

curl -X POST 'http://0.0.0.0:8126/test/session/responses/config/path' -d '{"path": "datadog/2/ASM_DATA/blocked_users/config", "msg": {"rules_data": []}}'

/test/trace_check/failures (GET)

Get Trace Check failures that occured. If a token is included, trace failures for only that session token are returned unless used in conjuction with return_all, which can be used to return all failures regardless of inputted token. This method returns a <Response 200> if no Trace Check failures are being returned and a <Response 400> if Trace Check failures are being returned. Trace Check failures are returned as a content type of text, with failure messages concatenated in the response body. Optionally, set the use_json query string parameter to true to return Trace Check failures as a JSON response in the following format:

response = { 
  "<FAILING_CHECK_NAME>" : ["<FAILURE_MESSAGE_1>", "<FAILURE_MESSAGE_2>"]
}

NOTE: To be used in combination with DD_POOL_TRACE_CHECK_FAILURES, or else failures will not be saved within Test-Agent memory and a <Response 200> will always be returned.

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?use_json=`

[optional] `?return_all=`

curl -X GET 'http://0.0.0.0:8126/test/trace_check/failures'

/test/trace_check/clear (GET)

Clear Trace Check failures that occured. If a token is included, trace failures for only that session token are cleared unless used in conjuction with clear_all. This argument can be used to clear all failures (regardless of inputted session token).

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?clear_all=`

curl -X GET 'http://0.0.0.0:8126/test/trace_check/clear'

/test/trace_check/summary (GET)

Get Trace Check summary results. If a token is included, returns summary results only for Trace Checks run during the session. The return_all optional query string parameter can be used to return all trace check results (regardless of inputted session token). The method returns Trace Check results in the following JSON format:

summary = { 
  "trace_content_length" : {
    "Passed_Checks": 10,
    "Failed_Checks": 0,
    "Skipped_Checks": 4,
  }  ...
}

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?return_all=`

curl -X GET 'http://0.0.0.0:8126/test/trace_check/summary'

/test/session/integrations (PUT)

Update information about the current tested integration.

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

curl -X PUT 'http://0.0.0.0:8126/test/session/integrations' -d '{"integration_name": [INTEGRATION_NAME], "integration_version": [INTEGRATION_VERSION],
"dependency_name": [DEPENDENCY_NAME], "tracer_language": [TRACER_LANGUAGE], "tracer_version": [TRACER_VERSION]}'

/test/integrations/tested_versions (GET)

Return a csv list of all tested integrations received by the agent. The format of returned data will be: tracer_language,tracer_version,integration_name,integration_version,dependency_name.

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

curl -X GET 'http://0.0.0.0:8126/test/integrations/tested_versions'

/v0.1/pipeline_stats

Mimics the pipeline_stats endpoint of the agent, but always returns OK, and logs a line everytime it's called.

/tracer_flare/v1

Mimics the tracer_flare endpoint of the agent. Returns OK if the flare contains the required form fields, otherwise 400.

Logs a line everytime it's called and stores the tracer flare details in the request under "_tracer_flare".

/test/session/tracerflares

Return all tracer-flares that have been received by the agent for the given session token.

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

Returns the tracer-flares in the following json format:

[
  {
    "source": "...",
    "case_id": "...",
    "email": "...",
    "hostname": "...",
    "flare_file": "...",
  }
]

flare_file is the base64 encoded content of the tracer-flare payload.

If there was an error parsing the tracer-flare form, that will be recorded under error.

/test/settings (POST)

Allows to change some settings on the fly. This endpoint takes a POST request with a json content listing the keys and values to apply.

{ 'key': value }

Supported keys:

trace_request_delay: sets a delay to apply to trace and telemetry requests

curl -X POST 'http://0.0.0.0:8126/test/settings' -d '{ "trace_request_delay": 5 }'

Development

Prerequisites

A Python version of 3.8 or above and riot are required. It is recommended to create and work out of a virtualenv:

python3.12 -m venv .venv
source .venv/bin/activate
pip install -e '.[testing]'

Running the tests

To run the tests (in Python 3.12):

riot run -p3.12 test

Note: if snapshots need to be (re)generated in the tests set the environment variable GENERATE_SNAPSHOTS=1.

GENERATE_SNAPSHOTS=1 riot run --pass-env -p3.12 test -k test_trace_missing_received

Linting and formatting

To lint, format and type-check the code:

riot run -s flake8
riot run -s fmt
riot run -s mypy

Docker

To build (and tag) the dockerfile:

docker build --tag testagent .

Run the tagged image:

docker run --rm -v ${PWD}/snaps:/snapshots --publish 8126:8126 agent

Release notes

This project follows semver and so bug fixes, breaking changes, new features, etc must be accompanied by a release note. To generate a release note:

riot run reno new <short-description-of-change>

document the changes in the generated file, remove the irrelevant sections and commit the release note with the change.

Releasing

Checkout the master branch and make sure it's up to date.

    git checkout master && git pull

Generate the release notes and use pandoc to format them for Github:

    riot run -s reno report --no-show-source | pandoc -f rst -t gfm --wrap=none

Copy the output into a new release: https://github.com/DataDog/dd-apm-test-agent/releases/new.

Enter a tag for the release (following semver) (eg. v1.1.3, v1.0.3, v1.2.0).
Use the tag without the v as the title.
Save the release as a draft and pass the link to someone else to give a quick review.
If all looks good hit publish

Name		Name	Last commit message	Last commit date
Latest commit History 240 Commits
.github/workflows		.github/workflows
ddapm_test_agent		ddapm_test_agent
releasenotes		releasenotes
tests		tests
.dockerignore		.dockerignore
.gitignore		.gitignore
Dockerfile		Dockerfile
LICENSE.BSD3		LICENSE.BSD3
LICENSE.apache2		LICENSE.apache2
README.md		README.md
pyproject.toml		pyproject.toml
riotfile.py		riotfile.py
setup.cfg		setup.cfg
setup.py		setup.py
test_deps.txt		test_deps.txt

License

Licenses found

DataDog/dd-apm-test-agent

Folders and files

Latest commit

History

Repository files navigation

Datadog APM test agent

Installation

Features

Trace invariant checks

Returning data

Helpful logging

Proxy

Snapshot testing

Snapshot output

Configuration

Command line

ddapm-test-agent

ddapm-test-agent-fmt

Environment Variables

HTTP API

/test/traces

[optional] ?trace_ids=

[optional] X-Datadog-Trace-Ids

/test/session/start

[optional] ?agent_sample_rate_by_service=

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/session/snapshot

[optional*] ?test_session_token=

[optional*] X-Datadog-Test-Session-Token

[optional] ?ignores=

[optional] ?dir=

[optional] ?file=

[optional] X-Datadog-Test-Snapshot-Filename

[optional] ?removes=

/test/session/requests

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/session/traces

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/session/stats

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/session/responses/config (POST)

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/session/responses/config/path (POST)

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/trace_check/failures (GET)

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

[optional] ?use_json=

[optional] ?return_all=

/test/trace_check/clear (GET)

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

[optional] ?clear_all=

/test/trace_check/summary (GET)

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

[optional] ?return_all=

/test/session/integrations (PUT)

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/integrations/tested_versions (GET)

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/v0.1/pipeline_stats

/tracer_flare/v1

/test/session/tracerflares

[optional] ?test_session_token=

[optional] X-Datadog-Test-Session-Token

/test/settings (POST)

Development

Prerequisites

Running the tests

[optional] `?trace_ids=`

[optional] `X-Datadog-Trace-Ids`

[optional] `?agent_sample_rate_by_service=`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional*] `?test_session_token=`

[optional*] `X-Datadog-Test-Session-Token`

[optional] `?ignores=`

[optional] `?dir=`

[optional] `?file=`

[optional] `X-Datadog-Test-Snapshot-Filename`

[optional] `?removes=`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?use_json=`

[optional] `?return_all=`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?clear_all=`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?return_all=`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

[optional] `?test_session_token=`

[optional] `X-Datadog-Test-Session-Token`

Packages