Traces downsampling policy

canonical · Sep 27, 2024 · ed4b71e · ed4b71e
1 parent 699e3d7
commit ed4b71e
Show file tree

Hide file tree

Showing 3 changed files with 128 additions and 0 deletions.
diff --git a/charmcraft-22.04.yaml b/charmcraft-22.04.yaml
@@ -175,3 +175,24 @@ config:
         even if there is no integration currently requesting it.
       type: boolean
       default: false
+    charm_traces_sampling_percentage:
+      description: >
+        This property defines the percentage of charm traces that are sent to tracing backend.
+        Setting it to 100 would mean all charm traces are kept, setting to 0 means charm traces
+        aren't sent to tracing backend at all.
+      type: float
+      default: 100.0
+    workload_traces_sampling_percentage:
+      description: >
+        This property defines the percentage of workload traces that are sent to tracing backend.
+        Setting it to 100 would mean all workload traces are kept, setting to 0 means workload traces
+        aren't sent to tracing backend at all.
+      type: float
+      default: 1.0
+    error_traces_sampling_percentage:
+      description: >
+        This property defines the percentage of error traces (regardless of the type) that are sent to tracing backend.
+        Setting it to 100 would mean all error traces are kept, setting to 0 means error traces
+        aren't sent to tracing backend at all.
+      type: float
+      default: 100.0
diff --git a/charmcraft-24.04.yaml b/charmcraft-24.04.yaml
@@ -140,3 +140,24 @@ config:
         even if there is no integration currently requesting it.
       type: boolean
       default: false
+    charm_traces_sampling_percentage:
+      description: >
+        This property defines the percentage of charm traces that are sent to tracing backend.
+        Setting it to 100 would mean all charm traces are kept, setting to 0 means charm traces
+        aren't sent to tracing backend at all.
+      type: float
+      default: 100.0
+    workload_traces_sampling_percentage:
+      description: >
+        This property defines the percentage of workload traces that are sent to tracing backend.
+        Setting it to 100 would mean all workload traces are kept, setting to 0 means workload traces
+        aren't sent to tracing backend at all.
+      type: float
+      default: 1.0
+    error_traces_sampling_percentage:
+      description: >
+        This property defines the percentage of error traces (regardless of the type) that are sent to tracing backend.
+        Setting it to 100 would mean all error traces are kept, setting to 0 means error traces
+        aren't sent to tracing backend at all.
+      type: float
+      default: 100.0
diff --git a/src/grafana_agent.py b/src/grafana_agent.py
@@ -868,6 +868,90 @@ def _receiver_config(protocol: str):
 
         return config
 
+
+    @property
+    def _tracing_sampling(self) -> Dict[str, Any]:
+        return {
+            "policies": [
+                {
+                    "name": "error-traces-policy",
+                    "type": "and",
+                    "and": {
+                        "and_sub_policy": [
+                            {
+                                "name": "trace-status-policy",
+                                "type": "status_code",
+                                "status_code": {"status_codes": ["ERROR"]},
+                            },
+                            {
+                                "name": "probabilistic-policy",
+                                "type": "probabilistic",
+                                "probabilistic": {
+                                    "sampling_percentage": self.config.get(
+                                        "error_traces_sampling_percentage"
+                                    )
+                                },
+                            },
+                        ]
+                    },
+                },
+                {
+                    "name": "charm-traces-policy",
+                    "type": "and",
+                    "and": {
+                        "and_sub_policy": [
+                            {
+                                "name": "service-name-policy",
+                                "type": "string_attribute",
+                                "string_attribute": {
+                                    "key": "service.name",
+                                    "values": [".+-charm"],
+                                    "enabled_regex_matching": True,
+                                },
+                            },
+                            {
+                                "name": "probabilistic-policy",
+                                "type": "probabilistic",
+                                "probabilistic": {
+                                    "sampling_percentage": self.config.get(
+                                        "charm_traces_sampling_percentage"
+                                    )
+                                },
+                            },
+                        ]
+                    },
+                },
+                {
+                    "name": "workload-traces-policy",
+                    "type": "and",
+                    "and": {
+                        "and_sub_policy": [
+                            {
+                                "name": "service-name-policy",
+                                "type": "string_attribute",
+                                "string_attribute": {
+                                    "key": "service.name",
+                                    "values": [".+-charm"],
+                                    "enabled_regex_matching": True,
+                                    "invert_match": True,
+                                },
+                            },
+                            {
+                                "name": "probabilistic-policy",
+                                "type": "probabilistic",
+                                "probabilistic": {
+                                    "sampling_percentage": self.config.get(
+                                        "workload_traces_sampling_percentage"
+                                    )
+                                },
+                            },
+                        ]
+                    },
+                },
+            ]
+        }
+
+
     @property
     def _tempo_config(self) -> Dict[str, Union[Any, List[Any]]]:
         """The tracing section of the config.
@@ -877,6 +961,7 @@ def _tempo_config(self) -> Dict[str, Union[Any, List[Any]]]:
         """
         endpoints = self._tempo_endpoints_with_tls()
         receivers = self._tracing_receivers
+        sampling = self._tracing_sampling
 
         if not receivers:
             # pushing a config with an empty receivers section will cause gagent to error out
@@ -888,6 +973,7 @@ def _tempo_config(self) -> Dict[str, Union[Any, List[Any]]]:
                     "name": "tempo",
                     "remote_write": endpoints,
                     "receivers": receivers,
+                    "tail_sampling": sampling,
                 }
             ]
         }