ref(feedback): add save_event_feedback metric and cleanup other ingest metrics (#97741)

- Adds a timer metric for the task feedback v2 ingest runs in -
`save_event_feedback`. (The v1 equivalent is
`ingest_consumer.process_userreport`)
- Rm all `sample_rate=1.0`s. Digging into metrics code, you can see this
is only used for internal metrics, which we don't use anymore.
- cleans up some unused metrics (see comments)

Author: aliu39

src/sentry/feedback/usecases/ingest/create_feedback.py

@@ -98,17 +98,19 @@ def fix_for_issue_platform(event_data: dict[str, Any]) -> dict[str, Any]:
 
     ret_event["contexts"] = event_data.get("contexts", {})
 
-    # TODO: remove this once feedback_ingest API deprecated
-    # as replay context will be filled in
+    # TODO: investigate if this can be removed. If the frontend looks in both
+    # feedback and replay context for replay_id, this might not be needed.
     if not event_data["contexts"].get("replay") and event_data["contexts"].get("feedback", {}).get(
         "replay_id"
     ):
-        # Temporary metric to confirm this behavior is no longer needed.
-        metrics.incr("feedback.create_feedback_issue.filled_missing_replay_context")
-
+        # This metric confirms this block is still entered.
+        metrics.incr(
+            "feedback.create_feedback_issue.filled_missing_replay_context",
+        )
         ret_event["contexts"]["replay"] = {
             "replay_id": event_data["contexts"].get("feedback", {}).get("replay_id")
         }
+
     ret_event["event_id"] = event_data["event_id"]
 
     ret_event["platform"] = event_data.get("platform", "other")
@@ -162,11 +164,7 @@ def validate_issue_platform_event_schema(event_data):
     The issue platform schema validation does not run in dev atm so we have to do the validation
     ourselves, or else our tests are not representative of what happens in prod.
     """
-    try:
-        jsonschema.validate(event_data, EVENT_PAYLOAD_SCHEMA)
-    except jsonschema.exceptions.ValidationError:
-        metrics.incr("feedback.create_feedback_issue.invalid_schema")
-        raise
+    jsonschema.validate(event_data, EVENT_PAYLOAD_SCHEMA)
 
 
 def should_filter_feedback(event: dict) -> tuple[bool, str | None]:
@@ -214,13 +212,6 @@ def create_feedback_issue(
     Returns the formatted event data that was sent to issue platform.
     """
 
-    metrics.incr(
-        "feedback.create_feedback_issue.entered",
-        tags={
-            "referrer": source.value,
-        },
-    )
-
     should_filter, filter_reason = should_filter_feedback(event)
     if should_filter:
         if filter_reason == "too_large":
@@ -279,13 +270,14 @@ def create_feedback_issue(
         except Exception:
             # until we have LLM error types ironed out, just catch all exceptions
             logger.exception("Error checking if message is spam", extra={"project_id": project.id})
+
+        # In DD we use is_spam = None to indicate spam failed.
         metrics.incr(
             "feedback.create_feedback_issue.spam_detection",
             tags={
                 "is_spam": is_message_spam,
                 "referrer": source.value,
             },
-            sample_rate=1.0,
         )
 
     # Prepare the data for issue platform processing and attach useful tags.
@@ -411,7 +403,6 @@ def create_feedback_issue(
             "referrer": source.value,
             "platform": project.platform,
         },
-        sample_rate=1.0,
     )
 
     track_outcome(

src/sentry/feedback/usecases/ingest/save_event_feedback.py

@@ -15,8 +15,8 @@
 
 def save_event_feedback(event_data: Mapping[str, Any], project_id: int):
     """Saves feedback given data in an event format. This function should only
-    be called by the new feedback consumer's ingest strategy, to process
-    feedback envelopes (feedback v2). It is currently instrumented as a task in
+    be called by the feedback consumer's ingest strategy, to process
+    event envelopes (feedback v2). It is currently called in a task in
     sentry.tasks.store.
 
     If the save is successful and the `associated_event_id` field is present,

src/sentry/feedback/usecases/label_generation.py

@@ -26,7 +26,7 @@ class LabelRequest(TypedDict):
 SEER_GENERATE_LABELS_URL = f"{settings.SEER_AUTOFIX_URL}/v1/automation/summarize/feedback/labels"
 
 
-@metrics.wraps("feedback.generate_labels", sample_rate=1.0)
+@metrics.wraps("feedback.generate_labels")
 def generate_labels(feedback_message: str, organization_id: int) -> list[str]:
     """
     Generate labels for a feedback message.

src/sentry/feedback/usecases/spam_detection.py

@@ -32,7 +32,7 @@ def make_input_prompt(message: str):
 **Classify:** """
 
 
-@metrics.wraps("feedback.spam_detection", sample_rate=1.0)
+@metrics.wraps("feedback.spam_detection")
 def is_spam(message: str):
     labeled_spam = False
     response = complete_prompt(

src/sentry/feedback/usecases/title_generation.py

@@ -26,17 +26,9 @@ class GenerateFeedbackTitleRequest(TypedDict):
 def should_get_ai_title(organization: Organization) -> bool:
     """Check if AI title generation should be used for the given organization."""
     if not features.has("organizations:gen-ai-features", organization):
-        metrics.incr(
-            "feedback.ai_title_generation.skipped",
-            tags={"reason": "gen_ai_disabled"},
-        )
         return False
 
     if not features.has("organizations:user-feedback-ai-titles", organization):
-        metrics.incr(
-            "feedback.ai_title_generation.skipped",
-            tags={"reason": "feedback_ai_titles_disabled"},
-        )
         return False
 
     return True
@@ -75,6 +67,7 @@ def format_feedback_title(title: str, max_words: int = 10) -> str:
     return title
 
 
+@metrics.wraps("feedback.ai_title_generation")
 def get_feedback_title_from_seer(feedback_message: str, organization_id: int) -> str | None:
     """
     Generate an AI-powered title for user feedback using Seer, or None if generation fails.
@@ -119,9 +112,6 @@ def get_feedback_title_from_seer(feedback_message: str, organization_id: int) ->
         )
         return None
 
-    metrics.incr(
-        "feedback.ai_title_generation.success",
-    )
     return title
 
 

src/sentry/tasks/store.py

@@ -717,6 +717,7 @@ def save_event_transaction(
         processing_deadline_duration=65,
     ),
 )
+@metrics.wraps("feedback_consumer.save_event_feedback_task")
 def save_event_feedback(
     cache_key: str | None = None,
     start_time: float | None = None,