feat: Update Influx schema with new fields/tags #29 (#30)

* feat: Update Influx schema with new fields/tags #29

* fix: change execution path to JSON array

---------

Co-authored-by: Aliaksandr Stsiapanay <aliaksandr_stsiapanay@epam.com>

aidial_analytics_realtime/analytics.py

@@ -69,6 +69,7 @@ def make_point(
     rates_calculator: RatesCalculator,
     parent_deployment: str | None,
     trace: dict | None,
+    execution_path: list | None,
 ):
     topic = None
     response_content = ""
@@ -94,8 +95,10 @@ def make_point(
             )
 
     price = Decimal(0)
+    deployment_price = Decimal(0)
     if usage is not None and usage.get("price") is not None:
         price = usage["price"]
+        deployment_price = usage.get("deployment_price", Decimal(0))
     else:
         price = rates_calculator.calculate_price(
             deployment, model, request_content, response_content, usage
@@ -106,6 +109,10 @@ def make_point(
         .tag("model", model)
         .tag("deployment", deployment)
         .tag("parent_deployment", to_string(parent_deployment))
+        .tag(
+            "execution_path",
+            "undefined" if not execution_path else "/".join(execution_path),
+        )
         .tag("trace_id", "undefined" if not trace else trace["trace_id"])
         .tag(
             "core_span_id", "undefined" if not trace else trace["core_span_id"]
@@ -134,6 +141,7 @@ def make_point(
         )
         .field("user_hash", to_string(user_hash))
         .field("price", price)
+        .field("deployment_price", deployment_price)
         .field(
             "number_request_messages",
             len(request["messages"])
@@ -198,6 +206,7 @@ async def on_message(
     token_usage: dict | None,
     parent_deployment: str | None,
     trace: dict | None,
+    execution_path: list | None,
 ):
     logger.info(f"Chat completion response length {len(response)}")
 
@@ -220,6 +229,7 @@ async def on_message(
             rates_calculator,
             parent_deployment,
             trace,
+            execution_path,
         )
         await influx_writer(point)
     elif len(usage_per_model) == 0:
@@ -240,6 +250,7 @@ async def on_message(
             rates_calculator,
             parent_deployment,
             trace,
+            execution_path,
         )
         await influx_writer(point)
     else:
@@ -260,6 +271,7 @@ async def on_message(
             rates_calculator,
             parent_deployment,
             trace,
+            execution_path,
         )
         await influx_writer(point)
 
@@ -281,5 +293,6 @@ async def on_message(
                 rates_calculator,
                 parent_deployment,
                 trace,
+                execution_path,
             )
             await influx_writer(point)

aidial_analytics_realtime/app.py

@@ -70,6 +70,7 @@ async def on_chat_completion_message(
     token_usage: dict | None,
     parent_deployment: str | None,
     trace: dict | None,
+    execution_path: list | None,
 ):
     if response["status"] != "200":
         return
@@ -124,6 +125,7 @@ async def on_chat_completion_message(
         token_usage,
         parent_deployment,
         trace,
+        execution_path,
     )
 
 
@@ -143,6 +145,7 @@ async def on_embedding_message(
     token_usage: dict | None,
     parent_deployment: str | None,
     trace: dict | None,
+    execution_path: list | None,
 ):
     if response["status"] != "200":
         return
@@ -166,6 +169,7 @@ async def on_embedding_message(
         token_usage,
         parent_deployment,
         trace,
+        execution_path,
     )
 
 
@@ -191,6 +195,7 @@ async def on_log_message(
     token_usage = message.get("token_usage", None)
     trace = message.get("trace", None)
     parent_deployment = message.get("parent_deployment", None)
+    execution_path = message.get("execution_path", None)
 
     match = re.search(RATE_PATTERN, uri)
     if match:
@@ -215,6 +220,7 @@ async def on_log_message(
             token_usage,
             parent_deployment,
             trace,
+            execution_path,
         )
 
     match = re.search(EMBEDDING_PATTERN, uri)
@@ -236,6 +242,7 @@ async def on_log_message(
             token_usage,
             parent_deployment,
             trace,
+            execution_path,
         )
 
 

tests/test_app.py

@@ -89,8 +89,8 @@ def test_data_request():
     )
     assert response.status_code == 200
     assert write_api_mock.points == [
-        'analytics,core_parent_span_id=undefined,core_span_id=undefined,deployment=gpt-4,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=TestTopic,trace_id=undefined,upstream=undefined chat_id="chat-1",completion_tokens=189i,number_request_messages=2i,price=0,prompt_tokens=22i,user_hash="undefined" 1692214959997000000',
-        'analytics,core_parent_span_id=undefined,core_span_id=undefined,deployment=gpt-4,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY-2,response_id=chatcmpl-2,title=undefined,topic=TestTopic,trace_id=undefined,upstream=undefined chat_id="chat-2",completion_tokens=189i,number_request_messages=2i,price=0,prompt_tokens=22i,user_hash="undefined" 1700796820390000000',
+        'analytics,core_parent_span_id=undefined,core_span_id=undefined,deployment=gpt-4,execution_path=undefined,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=TestTopic,trace_id=undefined,upstream=undefined chat_id="chat-1",completion_tokens=189i,deployment_price=0,number_request_messages=2i,price=0,prompt_tokens=22i,user_hash="undefined" 1692214959997000000',
+        'analytics,core_parent_span_id=undefined,core_span_id=undefined,deployment=gpt-4,execution_path=undefined,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY-2,response_id=chatcmpl-2,title=undefined,topic=TestTopic,trace_id=undefined,upstream=undefined chat_id="chat-2",completion_tokens=189i,deployment_price=0,number_request_messages=2i,price=0,prompt_tokens=22i,user_hash="undefined" 1700796820390000000',
     ]
 
 
@@ -125,6 +125,7 @@ def test_data_request_with_new_format():
                             "core_span_id": "9ade2b6fef0a716d",
                             "core_parent_span_id": "20e7e64715abbe97",
                         },
+                        "execution_path": ["a", "b", "c"],
                         "request": {
                             "protocol": "HTTP/1.1",
                             "method": "POST",
@@ -167,6 +168,7 @@ def test_data_request_with_new_format():
                             "trace_id": "5dca3d6ed5d22b6ab574f27a6ab5ec14",
                             "core_span_id": "20e7e64715abbe97",
                         },
+                        "execution_path": ["a", "b", "c"],
                         "request": {
                             "protocol": "HTTP/1.1",
                             "method": "POST",
@@ -197,6 +199,6 @@ def test_data_request_with_new_format():
     )
     assert response.status_code == 200
     assert write_api_mock.points == [
-        'analytics,core_parent_span_id=20e7e64715abbe97,core_span_id=9ade2b6fef0a716d,deployment=gpt-4,language=undefined,model=gpt-4,parent_deployment=assistant,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=TestTopic,trace_id=5dca3d6ed5d22b6ab574f27a6ab5ec14,upstream=undefined chat_id="chat-1",completion_tokens=40i,number_request_messages=2i,price=0.001,prompt_tokens=30i,user_hash="undefined" 1692214959997000000',
-        'analytics,core_parent_span_id=undefined,core_span_id=20e7e64715abbe97,deployment=gpt-4,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY-2,response_id=chatcmpl-2,title=undefined,topic=TestTopic,trace_id=5dca3d6ed5d22b6ab574f27a6ab5ec14,upstream=undefined chat_id="chat-2",completion_tokens=40i,number_request_messages=2i,price=0.005,prompt_tokens=30i,user_hash="undefined" 1700796820390000000',
+        'analytics,core_parent_span_id=20e7e64715abbe97,core_span_id=9ade2b6fef0a716d,deployment=gpt-4,execution_path=a/b/c,language=undefined,model=gpt-4,parent_deployment=assistant,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=TestTopic,trace_id=5dca3d6ed5d22b6ab574f27a6ab5ec14,upstream=undefined chat_id="chat-1",completion_tokens=40i,deployment_price=0.001,number_request_messages=2i,price=0.001,prompt_tokens=30i,user_hash="undefined" 1692214959997000000',
+        'analytics,core_parent_span_id=undefined,core_span_id=20e7e64715abbe97,deployment=gpt-4,execution_path=a/b/c,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY-2,response_id=chatcmpl-2,title=undefined,topic=TestTopic,trace_id=5dca3d6ed5d22b6ab574f27a6ab5ec14,upstream=undefined chat_id="chat-2",completion_tokens=40i,deployment_price=0,number_request_messages=2i,price=0.005,prompt_tokens=30i,user_hash="undefined" 1700796820390000000',
     ]

tests/test_huggingface_model.py

@@ -56,7 +56,7 @@ def test_data_request():
     )
     assert response.status_code == 200
     assert write_api_mock.points == [
-        'analytics,core_parent_span_id=undefined,core_span_id=undefined,deployment=gpt-4,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=4_letter_sincerely_regards_email,trace_id=undefined,upstream=undefined chat_id="chat-1",completion_tokens=189i,number_request_messages=2i,price=0,prompt_tokens=22i,user_hash="undefined" 1692214959997000000'
+        'analytics,core_parent_span_id=undefined,core_span_id=undefined,deployment=gpt-4,execution_path=undefined,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=4_letter_sincerely_regards_email,trace_id=undefined,upstream=undefined chat_id="chat-1",completion_tokens=189i,deployment_price=0,number_request_messages=2i,price=0,prompt_tokens=22i,user_hash="undefined" 1692214959997000000'
     ]
 
 
@@ -90,6 +90,7 @@ def test_data_request_with_new_format():
                             "core_span_id": "9ade2b6fef0a716d",
                             "core_parent_span_id": "20e7e64715abbe97",
                         },
+                        "execution_path": ["a", "b", "c"],
                         "request": {
                             "protocol": "HTTP/1.1",
                             "method": "POST",
@@ -120,5 +121,5 @@ def test_data_request_with_new_format():
     )
     assert response.status_code == 200
     assert write_api_mock.points == [
-        'analytics,core_parent_span_id=20e7e64715abbe97,core_span_id=9ade2b6fef0a716d,deployment=gpt-4,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=4_letter_sincerely_regards_email,trace_id=5dca3d6ed5d22b6ab574f27a6ab5ec14,upstream=undefined chat_id="chat-1",completion_tokens=40i,number_request_messages=2i,price=0.001,prompt_tokens=30i,user_hash="undefined" 1692214959997000000'
+        'analytics,core_parent_span_id=20e7e64715abbe97,core_span_id=9ade2b6fef0a716d,deployment=gpt-4,execution_path=a/b/c,language=undefined,model=gpt-4,parent_deployment=undefined,project_id=PROJECT-KEY,response_id=chatcmpl-1,title=undefined,topic=4_letter_sincerely_regards_email,trace_id=5dca3d6ed5d22b6ab574f27a6ab5ec14,upstream=undefined chat_id="chat-1",completion_tokens=40i,deployment_price=0.001,number_request_messages=2i,price=0.001,prompt_tokens=30i,user_hash="undefined" 1692214959997000000'
     ]