From 858cacdcfe9d702c9f7ef1ecde83b098c91b2891 Mon Sep 17 00:00:00 2001
From: Neel Shah <neel.shah@sentry.io>
Date: Wed, 18 Dec 2024 20:41:55 +0530
Subject: [PATCH] Implement span limits on span processor (#3881)

---
 sentry_sdk/integrations/arq.py                |  2 --
 sentry_sdk/integrations/logging.py            |  7 ++++++-
 .../opentelemetry/span_processor.py           | 19 ++++++++++++++++++-
 tests/tracing/test_misc.py                    | 19 +++++++++++++++++++
 4 files changed, 43 insertions(+), 4 deletions(-)

diff --git a/sentry_sdk/integrations/arq.py b/sentry_sdk/integrations/arq.py
index c26db4520c..0f42050cf2 100644
--- a/sentry_sdk/integrations/arq.py
+++ b/sentry_sdk/integrations/arq.py
@@ -1,7 +1,5 @@
 import sys
 
-from opentelemetry.trace.status import StatusCode
-
 import sentry_sdk
 from sentry_sdk.consts import OP, SPANSTATUS
 from sentry_sdk.integrations import DidNotEnable, Integration
diff --git a/sentry_sdk/integrations/logging.py b/sentry_sdk/integrations/logging.py
index 314780cabd..52c56a8e60 100644
--- a/sentry_sdk/integrations/logging.py
+++ b/sentry_sdk/integrations/logging.py
@@ -40,7 +40,12 @@
 # Note: Ignoring by logger name here is better than mucking with thread-locals.
 # We do not necessarily know whether thread-locals work 100% correctly in the user's environment.
 _IGNORED_LOGGERS = set(
-    ["sentry_sdk.errors", "urllib3.connectionpool", "urllib3.connection", "opentelemetry.*"]
+    [
+        "sentry_sdk.errors",
+        "urllib3.connectionpool",
+        "urllib3.connection",
+        "opentelemetry.*",
+    ]
 )
 
 
diff --git a/sentry_sdk/integrations/opentelemetry/span_processor.py b/sentry_sdk/integrations/opentelemetry/span_processor.py
index 0b4c3387df..42ad32a5ea 100644
--- a/sentry_sdk/integrations/opentelemetry/span_processor.py
+++ b/sentry_sdk/integrations/opentelemetry/span_processor.py
@@ -43,6 +43,9 @@
     from sentry_sdk._types import Event
 
 
+DEFAULT_MAX_SPANS = 1000
+
+
 class SentrySpanProcessor(SpanProcessor):
     """
     Converts OTel spans into Sentry spans so they can be sent to the Sentry backend.
@@ -79,7 +82,7 @@ def on_end(self, span):
             # if have a root span ending, we build a transaction and send it
             self._flush_root_span(span)
         else:
-            self._children_spans[span.parent.span_id].append(span)
+            self._append_child_span(span)
 
     # TODO-neel-potel not sure we need a clear like JS
     def shutdown(self):
@@ -150,6 +153,20 @@ def _flush_root_span(self, span):
 
         sentry_sdk.capture_event(transaction_event)
 
+    def _append_child_span(self, span):
+        # type: (ReadableSpan) -> None
+        if not span.parent:
+            return
+
+        max_spans = (
+            sentry_sdk.get_client().options["_experiments"].get("max_spans")
+            or DEFAULT_MAX_SPANS
+        )
+
+        children_spans = self._children_spans[span.parent.span_id]
+        if len(children_spans) < max_spans:
+            children_spans.append(span)
+
     def _collect_children(self, span):
         # type: (ReadableSpan) -> List[ReadableSpan]
         if not span.context:
diff --git a/tests/tracing/test_misc.py b/tests/tracing/test_misc.py
index 0d12acc617..dc5754cfd2 100644
--- a/tests/tracing/test_misc.py
+++ b/tests/tracing/test_misc.py
@@ -8,6 +8,25 @@
 from sentry_sdk.utils import Dsn
 
 
+def test_span_trimming(sentry_init, capture_events):
+    sentry_init(traces_sample_rate=1.0, _experiments={"max_spans": 3})
+    events = capture_events()
+
+    with start_span(name="hi"):
+        for i in range(10):
+            with start_span(op="foo{}".format(i)):
+                pass
+
+    (event,) = events
+
+    assert len(event["spans"]) == 3
+
+    span1, span2, span3 = event["spans"]
+    assert span1["op"] == "foo0"
+    assert span2["op"] == "foo1"
+    assert span3["op"] == "foo2"
+
+
 def test_transaction_naming(sentry_init, capture_events):
     sentry_init(traces_sample_rate=1.0)
     events = capture_events()