Merge branch '2.17' into juanjux/new-testrunner-for-flaky-tests-217

DataDog · Jan 15, 2025 · edb7a1c · edb7a1c
2 parents bfda6c4 + 3961c8d
commit edb7a1c
Show file tree

Hide file tree

Showing 9 changed files with 193 additions and 116 deletions.
diff --git a/.github/workflows/django-overhead-profile.yml b/.github/workflows/django-overhead-profile.yml
@@ -26,6 +26,7 @@ jobs:
       DD_PROFILING_ENABLED: "1"
       DD_PROFILING_STACK_V2_ENABLED: ${{ matrix.stack_v2 }}
       DD_PROFILING_OUTPUT_PPROF: ${{ github.workspace }}/prefix/artifacts/ddtrace_profile
+      DD_EXCEPTION_REPLAY_ENABLED: "1"
     defaults:
       run:
         working-directory: ddtrace

diff --git a/ddtrace/propagation/http.py b/ddtrace/propagation/http.py
@@ -1,3 +1,4 @@
+import itertools
 import re
 import sys
 from typing import Any  # noqa:F401
@@ -912,21 +913,23 @@ def _inject(span_context: Context, headers: Dict[str, str]) -> None:
         if not baggage_items:
             return
 
-        if len(baggage_items) > DD_TRACE_BAGGAGE_MAX_ITEMS:
-            log.warning("Baggage item limit exceeded")
-            return
-
         try:
-            header_value = ",".join(
-                f"{_BaggageHeader._encode_key(key)}={_BaggageHeader._encode_value(value)}"
-                for key, value in baggage_items
-            )
-
-            buf = bytes(header_value, "utf-8")
-            if len(buf) > DD_TRACE_BAGGAGE_MAX_BYTES:
-                log.warning("Baggage header size exceeded")
-                return
-
+            if len(baggage_items) > DD_TRACE_BAGGAGE_MAX_ITEMS:
+                log.warning("Baggage item limit exceeded, dropping excess items")
+                baggage_items = itertools.islice(baggage_items, DD_TRACE_BAGGAGE_MAX_ITEMS)  # type: ignore
+
+            encoded_items: List[str] = []
+            total_size = 0
+            for key, value in baggage_items:
+                item = f"{_BaggageHeader._encode_key(key)}={_BaggageHeader._encode_value(value)}"
+                item_size = len(item.encode("utf-8")) + (1 if encoded_items else 0)  # +1 for comma if not first item
+                if total_size + item_size > DD_TRACE_BAGGAGE_MAX_BYTES:
+                    log.warning("Baggage header size exceeded, dropping excess items")
+                    break  # stop adding items when size limit is reached
+                encoded_items.append(item)
+                total_size += item_size
+
+            header_value = ",".join(encoded_items)
             headers[_HTTP_HEADER_BAGGAGE] = header_value
 
         except Exception:

diff --git a/scripts/profiles/django-simple/k6-exc.js b/scripts/profiles/django-simple/k6-exc.js
@@ -0,0 +1,10 @@
+import http from 'k6/http';
+
+export const options = {
+    duration: '60s',
+    vus: 100,
+};
+
+export default function () {
+    const res = http.get('http://127.0.0.1:8080/polls/123/results/');
+}
diff --git a/scripts/profiles/django-simple/k6-load.js b/scripts/profiles/django-simple/k6-load.js
@@ -1,11 +1,10 @@
 import http from 'k6/http';
-import { sleep } from 'k6';
 
 export const options = {
     duration: '60s',
     vus: 100,
 };
 
 export default function () {
-    const res = http.get('http://127.0.0.1:8000/accounts/signup/');
+    const res = http.get('http://127.0.0.1:8080/polls/123/vote/');
 }
diff --git a/scripts/profiles/django-simple/run.sh b/scripts/profiles/django-simple/run.sh
@@ -9,59 +9,83 @@ AUSTIN_EXPOSURE=5  # sec
 
 test -f ${PREFIX}/gunicorn.pid && (kill -9 `cat ${PREFIX}/gunicorn.pid` ; sleep 3) || rm -f ${PREFIX}/gunicorn.pid
 pkill k6 || true
-test -d ${PREFIX}/artifacts && rm -rf ${PREFIX}/artifacts || mkdir -p ${PREFIX}/artifacts
+pkill -9 -f uwsgi || true
+test -d ${PREFIX}/artifacts && rm -rf ${PREFIX}/artifacts
+mkdir -p ${PREFIX}/artifacts
 
 sudo echo "sudo OK"
 
+sudo rm -f ${PREFIX}/uwsgi.pid
+
 function profile_with_load {
     name=${1}
+    scenario=${2}
+
+    echo "- profiling for ${name}"
 
     sleep 3
-    ${PREFIX}/k6*/k6 run --quiet scripts/profiles/django-simple/k6-load.js &
+    echo "Starting load"
+    ${PREFIX}/k6*/k6 run --quiet scripts/profiles/django-simple/k6-${scenario}.js &
         sleep 2
-        sudo `which austin` -bsCi ${AUSTIN_INTERVAL} -o ${PREFIX}/artifacts/${name}.mojo -p `cat ${PREFIX}/gunicorn.pid` -x ${AUSTIN_EXPOSURE}
-        LC_ALL=C sed -i 's|/home/runner/work/dd-trace-py/dd-trace-py/ddtrace/||g' ${PREFIX}/artifacts/${name}.mojo
+        echo "Attaching Austin to $(cat ${PREFIX}/uwsgi.pid)"
+        sudo `which austin` -bsCi ${AUSTIN_INTERVAL} -o ${PREFIX}/artifacts/${scenario}_${name}.mojo -p `cat ${PREFIX}/uwsgi.pid` -x ${AUSTIN_EXPOSURE}
+        LC_ALL=C sed -i 's|/home/runner/work/dd-trace-py/dd-trace-py/ddtrace/||g' ${PREFIX}/artifacts/${scenario}_${name}.mojo
+    echo "Stopping load"
     pkill k6
 }
 
 source ${PREFIX}/bin/activate
 
-export DJANGO_SETTINGS_MODULE="config.settings.production"
-export DJANGO_ALLOWED_HOSTS="127.0.0.1"
-export DJANGO_SECRET_KEY="SECRET_KEY"
 export DATABASE_URL="sqlite:///django.db"
+export DJANGO_ALLOWED_HOSTS="127.0.0.1"
+export DEVELOPMENT_MODE=True
 
 # Tag traces with HTTP headers to benchmark the related code
 export DD_TRACE_HEADER_TAGS="User-Agent:http.user_agent,Referer:http.referer,Content-Type:http.content_type,Etag:http.etag"
 
-# Baseline
-pushd ${PREFIX}/trace-examples/python/django/django-simple
-    gunicorn config.wsgi --pid ${PREFIX}/gunicorn.pid > /dev/null &
-    echo "Done"
-popd
-profile_with_load "baseline"
-kill $(cat ${PREFIX}/gunicorn.pid)
-
-pushd ${PREFIX}/trace-examples/python/django/django-simple
-    ddtrace-run gunicorn config.wsgi --pid ${PREFIX}/gunicorn.pid > /dev/null &
-popd
-profile_with_load "head"
-kill $(cat ${PREFIX}/gunicorn.pid)
-
-sudo chown -R $(id -u):$(id -g) ${PREFIX}/artifacts/*
-
-echo -n "Converting MOJO to Austin ... "
-mojo2austin ${PREFIX}/artifacts/head.mojo     ${PREFIX}/artifacts/head.austin.tmp
-mojo2austin ${PREFIX}/artifacts/baseline.mojo ${PREFIX}/artifacts/baseline.austin.tmp
-echo "[done]"
-
-echo -n "Diffing ... "
-python scripts/diff.py \
-    ${PREFIX}/artifacts/head.austin.tmp \
-    ${PREFIX}/artifacts/baseline.austin.tmp \
-    ${PREFIX}/artifacts/baseline_head.diff
-echo "[done]"
-
-rm ${PREFIX}/artifacts/*.austin.tmp
-
-head -n 25 ${PREFIX}/artifacts/baseline_head.diff.top
+
+function run_scenario {
+    scenario=${1}
+
+    echo "Running scenario ${scenario}"
+
+    # Baseline
+    pushd ${PREFIX}/trace-examples/python/django/sample-django
+        uwsgi --http :8080 --enable-threads --module mysite.wsgi --pidfile ${PREFIX}/uwsgi.pid 2> /dev/null &
+        echo "Done"
+    popd
+    profile_with_load "baseline" ${scenario}
+    echo "Stopping uwsgi"
+    uwsgi --stop ${PREFIX}/uwsgi.pid
+    pkill -9 -f uwsgi || true
+
+    pushd ${PREFIX}/trace-examples/python/django/sample-django
+        uwsgi --http :8080 --enable-threads --module mysite.wsgi --pidfile ${PREFIX}/uwsgi.pid --import=ddtrace.bootstrap.sitecustomize 2> /dev/null &
+    popd
+    profile_with_load "head" ${scenario}
+    echo "Stopping uwsgi"
+    uwsgi --stop ${PREFIX}/uwsgi.pid
+    pkill -9 -f uwsgi || true
+
+    sudo chown -R $(id -u):$(id -g) ${PREFIX}/artifacts/*
+
+    echo -n "Converting MOJO to Austin ... "
+    mojo2austin ${PREFIX}/artifacts/${scenario}_head.mojo     ${PREFIX}/artifacts/${scenario}_head.austin.tmp
+    mojo2austin ${PREFIX}/artifacts/${scenario}_baseline.mojo ${PREFIX}/artifacts/${scenario}_baseline.austin.tmp
+    echo "[done]"
+
+    echo -n "Diffing ... "
+    python scripts/diff.py \
+        ${PREFIX}/artifacts/${scenario}_head.austin.tmp \
+        ${PREFIX}/artifacts/${scenario}_baseline.austin.tmp \
+        ${PREFIX}/artifacts/${scenario}_baseline_head.diff
+    echo "[done]"
+
+    rm ${PREFIX}/artifacts/*.austin.tmp
+
+    head -n 25 ${PREFIX}/artifacts/${scenario}_baseline_head.diff.top
+}
+
+
+run_scenario "load"
+run_scenario "exc"
diff --git a/scripts/profiles/django-simple/setup.sh b/scripts/profiles/django-simple/setup.sh
@@ -6,9 +6,7 @@ PREFIX=${1}
 AUSTIN_VERSION="3.6"
 K6_VERSION="0.26.2"
 
-export DJANGO_SETTINGS_MODULE="config.settings.production"
 export DJANGO_ALLOWED_HOSTS="127.0.0.1"
-export DJANGO_SECRET_KEY="SECRET_KEY"
 export DATABASE_URL="sqlite:///django.db"
 
 # Clean up existing installation
@@ -26,12 +24,12 @@ source ${PREFIX}/bin/activate
 pip install pip --upgrade
 
 # Install the application
-git clone https://github.com/DataDog/trace-examples.git ${PREFIX}/trace-examples
+test -d ${PREFIX}/trace-examples || git clone -b sample-django --single-branch https://github.com/DataDog/trace-examples.git ${PREFIX}/trace-examples
 pushd ${PREFIX}/trace-examples/
-    git checkout origin/django-simple
-    pushd python/django/django-simple
-        pip install -r requirements/production.txt
+    pushd python/django/sample-django
+        pip install -r requirements.txt
         python manage.py migrate
+        python manage.py collectstatic
     popd
 popd
 

diff --git a/tests/contrib/botocore/test.py b/tests/contrib/botocore/test.py
@@ -23,6 +23,7 @@
 
 from ddtrace._trace._span_pointer import _SpanPointer
 from ddtrace._trace._span_pointer import _SpanPointerDirection
+from ddtrace._trace.utils_botocore import span_tags
 from tests.utils import get_128_bit_trace_id_from_headers
 
 
@@ -104,6 +105,12 @@ def setUp(self):
         super(BotocoreTest, self).setUp()
 
         Pin(service=self.TEST_SERVICE, tracer=self.tracer).onto(botocore.parsers.ResponseParser)
+        # Setting the validated flag to False ensures the redaction paths configurations are re-validated
+        # FIXME: Ensure AWSPayloadTagging._REQUEST_REDACTION_PATHS_DEFAULTS is always in sync with
+        # config.botocore.payload_tagging_request
+        # FIXME: Ensure AWSPayloadTagging._RESPONSE_REDACTION_PATHS_DEFAULTS is always in sync with
+        # config.botocore.payload_tagging_response
+        span_tags._PAYLOAD_TAGGER.validated = False
 
     def tearDown(self):
         super(BotocoreTest, self).tearDown()

diff --git a/tests/tracer/test_endpoint_config.py b/tests/tracer/test_endpoint_config.py
@@ -2,6 +2,7 @@
 
 from http.client import HTTPResponse
 from io import BytesIO
+import logging
 from unittest import mock
 
 from ddtrace.internal.http import HTTPConnection
@@ -84,87 +85,98 @@ def mock_pass(self, *args, **kwargs):
 
 
 def test_unset_config_endpoint(caplog):
-    caplog.set_level(10)
-    with override_env({"DD_TRACE_DEBUG": "true"}):
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"):
         assert fetch_config_from_endpoint() == {}
-    assert "Configuration endpoint not set. Skipping fetching configuration." in caplog.text
+    if caplog.text:
+        assert "Configuration endpoint not set. Skipping fetching configuration." in caplog.text
 
 
 def test_set_config_endpoint_enabled(caplog):
-    caplog.set_level(10)
-    with override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80", "DD_TRACE_DEBUG": "true"}), mock.patch.object(
-        HTTPConnection, "connect", new=mock_pass
-    ), mock.patch.object(HTTPConnection, "send", new=mock_pass), mock.patch.object(
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"), override_env(
+        {"_DD_CONFIG_ENDPOINT": "http://localhost:80"}
+    ), mock.patch.object(HTTPConnection, "connect", new=mock_pass), mock.patch.object(
+        HTTPConnection, "send", new=mock_pass
+    ), mock.patch.object(
         HTTPConnection, "getresponse", new=mock_getresponse_enabled
     ):
         assert fetch_config_from_endpoint() == {"dd_iast_enabled": True}
-    assert "Configuration endpoint not set. Skipping fetching configuration." not in caplog.text
-    assert "Failed to fetch configuration from endpoint" not in caplog.text
+    if caplog.text:
+        assert "Configuration endpoint not set. Skipping fetching configuration." not in caplog.text
+        assert "Failed to fetch configuration from endpoint" not in caplog.text
 
 
 def test_set_config_endpoint_500(caplog):
-    caplog.set_level(10)
-    with override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80"}), mock.patch.object(
-        HTTPConnection, "connect", new=mock_pass
-    ), mock.patch.object(HTTPConnection, "send", new=mock_pass), mock.patch.object(
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"), override_env(
+        {"_DD_CONFIG_ENDPOINT": "http://localhost:80"}
+    ), mock.patch.object(HTTPConnection, "connect", new=mock_pass), mock.patch.object(
+        HTTPConnection, "send", new=mock_pass
+    ), mock.patch.object(
         HTTPConnection, "getresponse", new=mock_getresponse_500
     ):
         assert fetch_config_from_endpoint() == {}
-    assert "Failed to fetch configuration from endpoint" in caplog.text
-    assert "RetryError: Response(status=500" in caplog.text
+    if caplog.text:
+        assert "Failed to fetch configuration from endpoint" in caplog.text
+        assert "RetryError: Response(status=500" in caplog.text
 
 
 def test_set_config_endpoint_403(caplog):
-    caplog.set_level(10)
-    with override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80"}), mock.patch.object(
-        HTTPConnection, "connect", new=mock_pass
-    ), mock.patch.object(HTTPConnection, "send", new=mock_pass), mock.patch.object(
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"), override_env(
+        {"_DD_CONFIG_ENDPOINT": "http://localhost:80"}
+    ), mock.patch.object(HTTPConnection, "connect", new=mock_pass), mock.patch.object(
+        HTTPConnection, "send", new=mock_pass
+    ), mock.patch.object(
         HTTPConnection, "getresponse", new=mock_getresponse_403
     ):
         assert fetch_config_from_endpoint() == {}
-    assert "Failed to fetch configuration from endpoint" in caplog.text
-    assert "RetryError: Response(status=403" in caplog.text
+    if caplog.text:
+        assert "Failed to fetch configuration from endpoint" in caplog.text
+        assert "RetryError: Response(status=403" in caplog.text
 
 
 def test_set_config_endpoint_malformed(caplog):
-    caplog.set_level(10)
-    with override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80"}), mock.patch.object(
-        HTTPConnection, "connect", new=mock_pass
-    ), mock.patch.object(HTTPConnection, "send", new=mock_pass), mock.patch.object(
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"), override_env(
+        {"_DD_CONFIG_ENDPOINT": "http://localhost:80"}
+    ), mock.patch.object(HTTPConnection, "connect", new=mock_pass), mock.patch.object(
+        HTTPConnection, "send", new=mock_pass
+    ), mock.patch.object(
         HTTPConnection, "getresponse", new=mock_getresponse_malformed
     ):
         assert fetch_config_from_endpoint() == {}
-    assert "Expecting property name enclosed in double quotes" in caplog.text
+    if caplog.text:
+        assert "Unable to parse Datadog Agent JSON" in caplog.text
 
 
 def test_set_config_endpoint_connection_refused(caplog):
-    caplog.set_level(10)
-    with override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80"}):
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"), override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80"}):
         assert fetch_config_from_endpoint() == {}
-    assert "Failed to fetch configuration from endpoint" in caplog.text
-    assert any(
-        message in caplog.text for message in ("Connection refused", "Address family not supported by protocol")
-    ), "None of the expected connection error log messages were found"
+
+    if caplog.text:
+        assert "Failed to fetch configuration from endpoint" in caplog.text
+        assert any(
+            message in caplog.text for message in ("Connection refused", "Address family not supported by protocol")
+        ), "None of the expected connection error log messages were found"
 
 
 def test_set_config_endpoint_timeout_error(caplog):
-    caplog.set_level(10)
-    with override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80", "DD_TRACE_DEBUG": "true"}), mock.patch(
-        "ddtrace.internal.utils.http.get_connection", side_effect=TimeoutError
-    ):
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"), override_env(
+        {"_DD_CONFIG_ENDPOINT": "http://localhost:80"}
+    ), mock.patch("ddtrace.internal.utils.http.get_connection", side_effect=TimeoutError):
         assert fetch_config_from_endpoint() == {}
-    assert "Configuration endpoint not set. Skipping fetching configuration." not in caplog.text
-    assert "Failed to fetch configuration from endpoint" in caplog.text
-    assert any(
-        message in caplog.text for message in ("Connection refused", "Address family not supported by protocol")
-    ), "None of the expected connection error log messages were found"
+
+    if caplog.text:
+        assert "Configuration endpoint not set. Skipping fetching configuration." not in caplog.text
+        assert "Failed to fetch configuration from endpoint" in caplog.text
+        assert any(
+            message in caplog.text for message in ("Connection refused", "Address family not supported by protocol")
+        ), "None of the expected connection error log messages were found"
 
 
 def test_set_config_endpoint_retries(caplog):
-    caplog.set_level(10)
-    with override_env({"_DD_CONFIG_ENDPOINT": "http://localhost:80", "DD_TRACE_DEBUG": "true"}), mock.patch.object(
-        HTTPConnection, "connect", new=mock_pass
-    ), mock.patch.object(HTTPConnection, "send", new=mock_pass), mock.patch.object(
+    with caplog.at_level(logging.DEBUG, logger="ddtrace"), override_env(
+        {"_DD_CONFIG_ENDPOINT": "http://localhost:80"}
+    ), mock.patch.object(HTTPConnection, "connect", new=mock_pass), mock.patch.object(
+        HTTPConnection, "send", new=mock_pass
+    ), mock.patch.object(
         HTTPConnection, "getresponse", new=mock_getresponse_enabled_after_4_retries
     ), mock.patch(
         "ddtrace.settings.endpoint_config._get_retries", return_value=5