[Tracing] Update DD_TAGS tests to remove empty tag values

manifests/python.yml

@@ -776,7 +776,7 @@ tests/:
     test_config_consistency.py:
       Test_Config_Dogstatsd: missing_feature
       Test_Config_RateLimit: v2.15.0
-      Test_Config_Tags: missing_feature
+      Test_Config_Tags: v2.0.0
       Test_Config_TraceAgentURL: v2.0.0
       Test_Config_TraceEnabled: v2.12.2
       Test_Config_TraceLogDirectory: missing_feature

tests/parametric/test_config_consistency.py

@@ -288,25 +288,64 @@ def test_setting_trace_rate_limit(self, library_env, test_agent, test_library):
         ), "Expected at least one trace to be rate-limited with sampling priority -1."
 
 
+class TagTest:
+    def __init__(self, expected_span_tags, agent_host_tags):
+        self.expected_span_tags = expected_span_tags
+        self.agent_host_tags = agent_host_tags
+
+
+# Feedback 1: Only difference is comma-split. Can we detect by looking at host-tags with commas present in them?
+# Feedback 2: Tracer is doing the right thing to discard <KEY>:<VALUE> tags where KEY is empty
 tag_scenarios: dict = {
-    "key1:value1,key2:value2": [("key1", "value1"), ("key2", "value2")],
-    "key1:value1 key2:value2": [("key1", "value1"), ("key2", "value2")],
-    "env:test aKey:aVal bKey:bVal cKey:": [("env", "test"), ("aKey", "aVal"), ("bKey", "bVal"), ("cKey", "")],
-    "env:test,aKey:aVal,bKey:bVal,cKey:": [("env", "test"), ("aKey", "aVal"), ("bKey", "bVal"), ("cKey", "")],
-    "env:test,aKey:aVal bKey:bVal cKey:": [("env", "test"), ("aKey", "aVal bKey:bVal cKey:")],
-    "env:test     bKey :bVal dKey: dVal cKey:": [
-        ("env", "test"),
-        ("bKey", ""),
-        ("dKey", ""),
-        ("dVal", ""),
-        ("cKey", ""),
-    ],
-    "env :test, aKey : aVal bKey:bVal cKey:": [("env", "test"), ("aKey", "aVal bKey:bVal cKey:")],
-    "env:keyWithA:Semicolon bKey:bVal cKey": [("env", "keyWithA:Semicolon"), ("bKey", "bVal"), ("cKey", "")],
-    "env:keyWith:  , ,   Lots:Of:Semicolons ": [("env", "keyWith:"), ("Lots", "Of:Semicolons")],
-    "a:b,c,d": [("a", "b"), ("c", ""), ("d", "")],
-    "a,1": [("a", ""), ("1", "")],
-    "a:b:c:d": [("a", "b:c:d")],
+    "key1:value1,key2:value2": TagTest(  # See Feedback 1
+        expected_span_tags=[("key1", "value1"), ("key2", "value2")], agent_host_tags=[("key1:value1,key2:value2")]
+    ),
+    "key1:value1 key2:value2": TagTest(
+        expected_span_tags=[("key1", "value1"), ("key2", "value2")],  # OK
+        agent_host_tags=["key1:value1", "key2:value2"],
+    ),
+    "env:test aKey:aVal bKey:bVal cKey:": TagTest(
+        expected_span_tags=[("env", "test"), ("aKey", "aVal"), ("bKey", "bVal"), ("cKey", "")],  # OK
+        agent_host_tags=["env:test", "aKey:aVal", "bKey:bVal", "cKey:"],
+    ),
+    "env:test,aKey:aVal,bKey:bVal,cKey:": TagTest(  # See Feedback 1
+        expected_span_tags=[("env", "test"), ("aKey", "aVal"), ("bKey", "bVal"), ("cKey", "")],
+        agent_host_tags=["env:test,aKey:aVal,bKey:bVal,cKey:"],
+    ),
+    # New test case adopted from the above ^ to take the simple comma-separated list into a (comma+space)-separated list
+    "env:test ,aKey:aVal, bKey:bVal, cKey:": TagTest(  # Feedback: Can we recommend users update their DD_TAGS to be (comma+space)-separate in their ENV variables AND toggle a feature flag to trim the commas?
+        expected_span_tags=[("env", "test"), ("aKey", "aVal"), ("bKey", "bVal"), ("cKey", "")],
+        agent_host_tags=["env:test", ",aKey:aVal,", "bKey:bVal,", "cKey:"],
+    ),
+    "env:test,aKey:aVal bKey:bVal cKey:": TagTest(  # See Feedback 1
+        expected_span_tags=[("env", "test"), ("aKey", "aVal bKey:bVal cKey:")],
+        agent_host_tags=["env:test,aKey:aVal", "bKey:bVal", "cKey:"],
+    ),
+    "env:test     bKey :bVal dKey: dVal cKey:": TagTest(
+        expected_span_tags=[("env", "test"), ("bKey", ""), ("dKey", ""), ("dVal", ""), ("cKey", "")],  # See Feedback 2
+        agent_host_tags=["env:test", "bKey", ":bVal", "dKey:", "dVal", "cKey:"],
+    ),
+    "env :test, aKey : aVal bKey:bVal cKey:": TagTest(  # See Feedback 1 and Feedback 2
+        expected_span_tags=[("env", "test"), ("aKey", "aVal bKey:bVal cKey:")],
+        agent_host_tags=["env", ":", ":test,", "aKey", "aVal", "bKey:bVal", "cKey:"],
+    ),
+    "env:keyWithA:Semicolon bKey:bVal cKey": TagTest(  # OK
+        expected_span_tags=[("env", "keyWithA:Semicolon"), ("bKey", "bVal"), ("cKey", "")],
+        agent_host_tags=["env:keyWithA:Semicolon", "bKey:bVal", "cKey"],
+    ),
+    "env:keyWith:  , ,   Lots:Of:Semicolons ": TagTest(  # See Feedback 1
+        expected_span_tags=[("env", "keyWith:"), ("Lots", "Of:Semicolons")],
+        agent_host_tags=["env:keyWith:", ",", "Lots:Of:Semicolons"],
+    ),
+    "a:b,c,d": TagTest(  # See Feedback 1
+        expected_span_tags=[("a", "b"), ("c", ""), ("d", "")], agent_host_tags=["a:b,c,d"]
+    ),
+    "a,1": TagTest(  # See Feedback 1
+        expected_span_tags=[("a", ""), ("1", "")], agent_host_tags=["a,1"]
+    ),
+    "a:b:c:d": TagTest(  # OK
+        expected_span_tags=[("a", "b:c:d")], agent_host_tags=["a:b:c:d"]
+    ),
 }
 
 
@@ -315,14 +354,16 @@ def test_setting_trace_rate_limit(self, library_env, test_agent, test_library):
 class Test_Config_Tags:
     @parametrize("library_env", [{"DD_TAGS": key} for key in tag_scenarios.keys()])
     def test_comma_space_tag_separation(self, library_env, test_agent, test_library):
-        expected_local_tags = []
+        expected_span_tags = []
         if "DD_TAGS" in library_env:
-            expected_local_tags = tag_scenarios[library_env["DD_TAGS"]]
+            tagtest = tag_scenarios[library_env["DD_TAGS"]]
+            expected_span_tags = tagtest.expected_span_tags
+
         with test_library:
             with test_library.dd_start_span(name="sample_span"):
                 pass
         span = find_only_span(test_agent.wait_for_num_traces(1))
-        for k, v in expected_local_tags:
+        for k, v in expected_span_tags:
             assert k in span["meta"]
             assert span["meta"][k] == v
 

tests/test_semantic_conventions.py

@@ -3,9 +3,10 @@
 # Copyright 2021 Datadog, Inc.
 
 import re
+import pprint
 from urllib.parse import urlparse
 
-from utils import context, interfaces, bug, missing_feature, features, scenarios
+from utils import context, interfaces, bug, missing_feature, features, scenarios, weblog
 
 RUNTIME_LANGUAGE_MAP = {
     "nodejs": "javascript",
@@ -323,6 +324,25 @@ def validator(span):
         interfaces.library.validate_spans(validator=validator)
 
 
+@scenarios.default
+@features.tracing_configuration_consistency
+class Test_Agent_DDTags:
+    """Spans carry meta tags that were set in DD_TAGS agent environment"""
+
+    def setup_main(self):
+        self.r = weblog.get("/")
+
+    def test_main(self):
+        assert self.r.status_code == 200
+
+        spans = interfaces.agent.get_spans_list(self.r)
+        assert len(spans) == 1, "Agent received the incorrect amount of spans"
+
+        host_tags = interfaces.agent.get_host_tags()
+        pprint.pprint(host_tags["system"])
+        assert False
+
+
 @features.data_integrity
 @features.envoy_external_processing
 @scenarios.external_processing

utils/_context/containers.py

@@ -537,6 +537,7 @@ def __init__(self, host_log_folder, *, use_proxy=True, environment=None) -> None
                 "DD_APM_RECEIVER_PORT": self.apm_receiver_port,
                 "DD_DOGSTATSD_PORT": self.dogstatsd_port,
                 "DD_API_KEY": os.environ.get("DD_API_KEY", _FAKE_DD_API_KEY),
+                # To configure configure host tags, set the env variable "DD_TAGS" here
             }
         )
 

utils/interfaces/_agent.py

@@ -117,6 +117,16 @@ def add_traces_validation(self, validator, *, success_by_default=False):
             validator=validator, success_by_default=success_by_default, path_filters=r"/api/v0\.[1-9]+/traces"
         )
 
+    def get_host_tags(self):
+        for data in self.get_data(path_filters="/intake/"):
+            if "host-tags" not in data["request"]["content"]:
+                continue
+
+            if data["request"]["content"].get("host-tags"):
+                return data["request"]["content"]["host-tags"]
+
+        return {}
+
     def get_spans(self, request=None):
         """Attempts to fetch the spans the agent will submit to the backend.