Add support 4 partial granularity tracing

Author: hmstepanek

newrelic/api/transaction.py

@@ -637,6 +637,7 @@ def __exit__(self, exc, value, tb):
             trace_id=self.trace_id,
             loop_time=self._loop_time,
             root=root_node,
+            partial_granularity_sampled=hasattr(self, "partial_granularity_sampled"),
         )
 
         # Clear settings as we are all done and don't need it
@@ -1073,23 +1074,52 @@ def _make_sampling_decision(self):
             return
         priority = self._priority
         sampled = self._sampled
-        _logger.debug(
-            "Full granularity tracing is enabled. Asking if full granularity wants to sample. priority=%s, sampled=%s",
-            priority,
-            sampled,
-        )
-        computed_priority, computed_sampled = self._compute_sampled_and_priority(
-            priority,
-            sampled,
-            remote_parent_sampled_path="distributed_tracing.sampler.remote_parent_sampled",
-            remote_parent_sampled_setting=self.settings.distributed_tracing.sampler.remote_parent_sampled,
-            remote_parent_not_sampled_path="distributed_tracing.sampler.remote_parent_not_sampled",
-            remote_parent_not_sampled_setting=self.settings.distributed_tracing.sampler.remote_parent_not_sampled,
-        )
-        _logger.debug("Full granularity sampling decision was %s with priority=%s.", sampled, priority)
-        self._priority = computed_priority
-        self._sampled = computed_sampled
-        self._sampling_decision_made = True
+        # Compute sampling decision for full granularity.
+        if self.settings.distributed_tracing.sampler.full_granularity.enabled:
+            _logger.debug(
+                "Full granularity tracing is enabled. Asking if full granularity wants to sample. priority=%s, sampled=%s",
+                priority,
+                sampled,
+            )
+            computed_priority, computed_sampled = self._compute_sampled_and_priority(
+                priority,
+                sampled,
+                remote_parent_sampled_path="distributed_tracing.sampler.full_granularity.remote_parent_sampled",
+                remote_parent_sampled_setting=self.settings.distributed_tracing.sampler.full_granularity.remote_parent_sampled,
+                remote_parent_not_sampled_path="distributed_tracing.sampler.full_granularity.remote_parent_not_sampled",
+                remote_parent_not_sampled_setting=self.settings.distributed_tracing.sampler.full_granularity.remote_parent_not_sampled,
+            )
+            _logger.debug("Full granularity sampling decision was %s with priority=%s.", sampled, priority)
+            if computed_sampled or not self.settings.distributed_tracing.sampler.partial_granularity.enabled:
+                self._priority = computed_priority
+                self._sampled = computed_sampled
+                self._sampling_decision_made = True
+                return
+
+        # If full granularity is not going to sample, let partial granularity decide.
+        if self.settings.distributed_tracing.sampler.partial_granularity.enabled:
+            _logger.debug("Partial granularity tracing is enabled. Asking if partial granularity wants to sample.")
+            self._priority, self._sampled = self._compute_sampled_and_priority(
+                priority,
+                sampled,
+                remote_parent_sampled_path="distributed_tracing.sampler.partial_granularity.remote_parent_sampled",
+                remote_parent_sampled_setting=self.settings.distributed_tracing.sampler.partial_granularity.remote_parent_sampled,
+                remote_parent_not_sampled_path="distributed_tracing.sampler.partial_granularity.remote_parent_not_sampled",
+                remote_parent_not_sampled_setting=self.settings.distributed_tracing.sampler.partial_granularity.remote_parent_not_sampled,
+            )
+            _logger.debug(
+                "Partial granularity sampling decision was %s with priority=%s.", self._sampled, self._priority
+            )
+            self._sampling_decision_made = True
+            if self._sampled:
+                self.partial_granularity_sampled = True
+            return
+
+        # This is only reachable if both full and partial granularity tracing are off.
+        # Set priority=0 and do not sample. This enables DT headers to still be sent
+        # even if the trace is never sampled.
+        self._priority = 0
+        self._sampled = False
 
     def _freeze_path(self):
         if self._frozen_path is None:

newrelic/config.py

@@ -407,6 +407,13 @@ def _process_configuration(section):
     _process_setting(section, "distributed_tracing.sampler.adaptive_sampling_target", "getint", None)
     _process_setting(section, "distributed_tracing.sampler.remote_parent_sampled", "get", None)
     _process_setting(section, "distributed_tracing.sampler.remote_parent_not_sampled", "get", None)
+    _process_setting(section, "distributed_tracing.sampler.full_granularity.enabled", "getboolean", None)
+    _process_setting(section, "distributed_tracing.sampler.full_granularity.remote_parent_sampled", "get", None)
+    _process_setting(section, "distributed_tracing.sampler.full_granularity.remote_parent_not_sampled", "get", None)
+    _process_setting(section, "distributed_tracing.sampler.partial_granularity.enabled", "getboolean", None)
+    _process_setting(section, "distributed_tracing.sampler.partial_granularity.type", "get", None)
+    _process_setting(section, "distributed_tracing.sampler.partial_granularity.remote_parent_sampled", "get", None)
+    _process_setting(section, "distributed_tracing.sampler.partial_granularity.remote_parent_not_sampled", "get", None)
     _process_setting(section, "span_events.enabled", "getboolean", None)
     _process_setting(section, "span_events.max_samples_stored", "getint", None)
     _process_setting(section, "span_events.attributes.enabled", "getboolean", None)

newrelic/core/attribute.py

@@ -109,6 +109,23 @@
     "zeebe.client.resourceFile",
 }
 
+SPAN_ENTITY_RELATIONSHIP_ATTRIBUTES = {
+    "cloud.account.id",
+    "cloud.platform",
+    "cloud.region",
+    "cloud.resource_id",
+    "db.instance",
+    "db.system",
+    "http.url",
+    "messaging.destination.name",
+    "messaging.system",
+    "peer.hostname",
+    "server.address",
+    "server.port",
+    "span.kind",
+}
+
+
 MAX_NUM_USER_ATTRIBUTES = 128
 MAX_ATTRIBUTE_LENGTH = 255
 MAX_NUM_ML_USER_ATTRIBUTES = 64

newrelic/core/config.py

@@ -337,6 +337,14 @@ class DistributedTracingSamplerSettings(Settings):
     pass
 
 
+class DistributedTracingSamplerFullGranularitySettings(Settings):
+    pass
+
+
+class DistributedTracingSamplerPartialGranularitySettings(Settings):
+    pass
+
+
 class ServerlessModeSettings(Settings):
     pass
 
@@ -507,6 +515,8 @@ class EventHarvestConfigHarvestLimitSettings(Settings):
 _settings.debug = DebugSettings()
 _settings.distributed_tracing = DistributedTracingSettings()
 _settings.distributed_tracing.sampler = DistributedTracingSamplerSettings()
+_settings.distributed_tracing.sampler.full_granularity = DistributedTracingSamplerFullGranularitySettings()
+_settings.distributed_tracing.sampler.partial_granularity = DistributedTracingSamplerPartialGranularitySettings()
 _settings.error_collector = ErrorCollectorSettings()
 _settings.error_collector.attributes = ErrorCollectorAttributesSettings()
 _settings.event_harvest_config = EventHarvestConfigSettings()
@@ -845,6 +855,27 @@ def default_otlp_host(host):
 _settings.distributed_tracing.sampler.remote_parent_not_sampled = os.environ.get(
     "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_REMOTE_PARENT_NOT_SAMPLED", "default"
 )
+_settings.distributed_tracing.sampler.full_granularity.enabled = _environ_as_bool(
+    "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_FULL_GRANULARITY_ENABLED", default=True
+)
+_settings.distributed_tracing.sampler.full_granularity.remote_parent_sampled = os.environ.get(
+    "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_FULL_GRANULARITY_REMOTE_PARENT_SAMPLED", None
+)
+_settings.distributed_tracing.sampler.full_granularity.remote_parent_not_sampled = os.environ.get(
+    "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_FULL_GRANULARITY_REMOTE_PARENT_NOT_SAMPLED", None
+)
+_settings.distributed_tracing.sampler.partial_granularity.enabled = _environ_as_bool(
+    "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_PARTIAL_GRANULARITY_ENABLED", default=False
+)
+_settings.distributed_tracing.sampler.partial_granularity.type = os.environ.get(
+    "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_PARTIAL_GRANULARITY_TYPE", "essential"
+)
+_settings.distributed_tracing.sampler.partial_granularity.remote_parent_sampled = os.environ.get(
+    "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_PARTIAL_GRANULARITY_REMOTE_PARENT_SAMPLED", "default"
+)
+_settings.distributed_tracing.sampler.partial_granularity.remote_parent_not_sampled = os.environ.get(
+    "NEW_RELIC_DISTRIBUTED_TRACING_SAMPLER_PARTIAL_GRANULARITY_REMOTE_PARENT_NOT_SAMPLED", "default"
+)
 _settings.distributed_tracing.exclude_newrelic_header = False
 _settings.span_events.enabled = _environ_as_bool("NEW_RELIC_SPAN_EVENTS_ENABLED", default=True)
 _settings.event_harvest_config.harvest_limits.span_event_data = _environ_as_int(
@@ -1371,9 +1402,34 @@ def finalize_application_settings(server_side_config=None, settings=_settings):
 
     application_settings.attribute_filter = AttributeFilter(flatten_settings(application_settings))
 
+    simplify_distributed_tracing_sampler_granularity_settings(application_settings)
+
     return application_settings
 
 
+def simplify_distributed_tracing_sampler_granularity_settings(settings):
+    # Full granularity settings may appear under:
+    # * `distributed_tracing.sampler`
+    # * `distributed_tracing.sampler.full_granularity`
+    # The `distributed_tracing.sampler.full_granularity` path takes precedence.
+    # To simplify logic in the code that uses these settings, store the values that
+    # should be used at the `distributed_tracing.sampler.full_granularity` path.
+    if not settings.distributed_tracing.sampler.full_granularity.remote_parent_sampled:
+        settings.distributed_tracing.sampler.full_granularity.remote_parent_sampled = (
+            settings.distributed_tracing.sampler.remote_parent_sampled
+        )
+    if not settings.distributed_tracing.sampler.full_granularity.remote_parent_not_sampled:
+        settings.distributed_tracing.sampler.full_granularity.remote_parent_not_sampled = (
+            settings.distributed_tracing.sampler.remote_parent_not_sampled
+        )
+    # Partial granularity tracing is not available in infinite tracing mode.
+    if settings.infinite_tracing.enabled and settings.distributed_tracing.sampler.partial_granularity.enabled:
+        _logger.warning(
+            "Improper configuration. Infinite tracing cannot be enabled at the same time as partial granularity tracing. Setting distributed_tracing.sampler.partial_granularity.enabled=False."
+        )
+        settings.distributed_tracing.sampler.partial_granularity.enabled = False
+
+
 def _remove_ignored_configs(server_settings):
     if not server_settings.get("agent_config"):
         return server_settings

newrelic/core/data_collector.py

@@ -117,7 +117,14 @@ def send_ml_events(self, sampling_info, custom_event_data):
 
     def send_span_events(self, sampling_info, span_event_data):
         """Called to submit sample set for span events."""
-
+        # TODO: remove this later after list types are suported.
+        for span_event in span_event_data:
+            try:
+                ids = span_event[1].get("nr.ids")
+                if ids:
+                    span_event[1]["nr.ids"] = ",".join(ids)
+            except:
+                pass
         payload = (self.agent_run_id, sampling_info, span_event_data)
         return self._protocol.send("span_event_data", payload)
 

newrelic/core/database_node.py

@@ -279,7 +279,15 @@ def trace_node(self, stats, root, connections):
             start_time=start_time, end_time=end_time, name=name, params=params, children=children, label=None
         )
 
-    def span_event(self, settings, base_attrs=None, parent_guid=None, attr_class=dict):
+    def span_event(
+        self,
+        settings,
+        base_attrs=None,
+        parent_guid=None,
+        attr_class=dict,
+        partial_granularity_sampled=False,
+        ct_exit_spans=None,
+    ):
         sql = self.formatted
 
         if sql:
@@ -288,4 +296,11 @@ def span_event(self, settings, base_attrs=None, parent_guid=None, attr_class=dic
 
         self.agent_attributes["db.statement"] = sql
 
-        return super().span_event(settings, base_attrs=base_attrs, parent_guid=parent_guid, attr_class=attr_class)
+        return super().span_event(
+            settings,
+            base_attrs=base_attrs,
+            parent_guid=parent_guid,
+            attr_class=attr_class,
+            partial_granularity_sampled=partial_granularity_sampled,
+            ct_exit_spans=ct_exit_spans,
+        )

newrelic/core/external_node.py

@@ -169,7 +169,15 @@ def trace_node(self, stats, root, connections):
             start_time=start_time, end_time=end_time, name=name, params=params, children=children, label=None
         )
 
-    def span_event(self, settings, base_attrs=None, parent_guid=None, attr_class=dict):
+    def span_event(
+        self,
+        settings,
+        base_attrs=None,
+        parent_guid=None,
+        attr_class=dict,
+        partial_granularity_sampled=False,
+        ct_exit_spans=None,
+    ):
         self.agent_attributes["http.url"] = self.http_url
 
         i_attrs = (base_attrs and base_attrs.copy()) or attr_class()
@@ -180,4 +188,11 @@ def span_event(self, settings, base_attrs=None, parent_guid=None, attr_class=dic
         if self.method:
             _, i_attrs["http.method"] = attribute.process_user_attribute("http.method", self.method)
 
-        return super().span_event(settings, base_attrs=i_attrs, parent_guid=parent_guid, attr_class=attr_class)
+        return super().span_event(
+            settings,
+            base_attrs=i_attrs,
+            parent_guid=parent_guid,
+            attr_class=attr_class,
+            partial_granularity_sampled=partial_granularity_sampled,
+            ct_exit_spans=ct_exit_spans,
+        )

newrelic/core/function_node.py

@@ -114,8 +114,23 @@ def trace_node(self, stats, root, connections):
             start_time=start_time, end_time=end_time, name=name, params=params, children=children, label=self.label
         )
 
-    def span_event(self, settings, base_attrs=None, parent_guid=None, attr_class=dict):
+    def span_event(
+        self,
+        settings,
+        base_attrs=None,
+        parent_guid=None,
+        attr_class=dict,
+        partial_granularity_sampled=False,
+        ct_exit_spans=None,
+    ):
         i_attrs = (base_attrs and base_attrs.copy()) or attr_class()
         i_attrs["name"] = f"{self.group}/{self.name}"
 
-        return super().span_event(settings, base_attrs=i_attrs, parent_guid=parent_guid, attr_class=attr_class)
+        return super().span_event(
+            settings,
+            base_attrs=i_attrs,
+            parent_guid=parent_guid,
+            attr_class=attr_class,
+            partial_granularity_sampled=partial_granularity_sampled,
+            ct_exit_spans=ct_exit_spans,
+        )

newrelic/core/loop_node.py

@@ -79,8 +79,23 @@ def trace_node(self, stats, root, connections):
             start_time=start_time, end_time=end_time, name=name, params=params, children=children, label=None
         )
 
-    def span_event(self, settings, base_attrs=None, parent_guid=None, attr_class=dict):
+    def span_event(
+        self,
+        settings,
+        base_attrs=None,
+        parent_guid=None,
+        attr_class=dict,
+        partial_granularity_sampled=False,
+        ct_exit_spans=None,
+    ):
         i_attrs = (base_attrs and base_attrs.copy()) or attr_class()
         i_attrs["name"] = f"EventLoop/Wait/{self.name}"
 
-        return super().span_event(settings, base_attrs=i_attrs, parent_guid=parent_guid, attr_class=attr_class)
+        return super().span_event(
+            settings,
+            base_attrs=i_attrs,
+            parent_guid=parent_guid,
+            attr_class=attr_class,
+            partial_granularity_sampled=partial_granularity_sampled,
+            ct_exit_spans=ct_exit_spans,
+        )