From 5e2fac8c3f0f704d9f126fb81d3414d4c98803d1 Mon Sep 17 00:00:00 2001
From: Cagri Yonca <cagri@ibm.com>
Date: Sat, 23 Aug 2025 16:03:08 +0200
Subject: [PATCH] fix(urllib3): ignore internal urllib3 span creation

Signed-off-by: Cagri Yonca <cagri@ibm.com>
---
 src/instana/instrumentation/urllib3.py | 24 ++++++++++++++--
 tests/clients/test_urllib3.py          | 38 ++++++++++++++++++++++++++
 2 files changed, 60 insertions(+), 2 deletions(-)

diff --git a/src/instana/instrumentation/urllib3.py b/src/instana/instrumentation/urllib3.py
index 4536d2be..52d3e9c8 100644
--- a/src/instana/instrumentation/urllib3.py
+++ b/src/instana/instrumentation/urllib3.py
@@ -11,7 +11,11 @@
 from instana.propagators.format import Format
 from instana.singletons import agent
 from instana.util.secrets import strip_secrets_from_query
-from instana.util.traceutils import get_tracer_tuple, tracing_is_off, extract_custom_headers
+from instana.util.traceutils import (
+    get_tracer_tuple,
+    tracing_is_off,
+    extract_custom_headers,
+)
 
 if TYPE_CHECKING:
     from instana.span.span import InstanaSpan
@@ -91,7 +95,23 @@ def urlopen_with_instana(
         tracer, parent_span, span_name = get_tracer_tuple()
 
         # If we're not tracing, just return; boto3 has it's own visibility
-        if tracing_is_off() or (span_name == "boto3"):
+        # Also, skip creating spans for internal Instana calls when
+        # 'com.instana' appears in either the full URL, the path argument,
+        # or the connection host.
+        request_url_or_path = (
+            kwargs.get("request_url")
+            or kwargs.get("url")
+            or (args[1] if len(args) >= 2 else "")
+            or ""
+        )
+        host = getattr(instance, "host", "") or ""
+
+        if (
+            tracing_is_off()
+            or span_name == "boto3"
+            or "com.instana" in request_url_or_path
+            or "com.instana" in host
+        ):
             return wrapped(*args, **kwargs)
 
         parent_context = parent_span.get_span_context() if parent_span else None
diff --git a/tests/clients/test_urllib3.py b/tests/clients/test_urllib3.py
index 62b07d49..6c5fc318 100644
--- a/tests/clients/test_urllib3.py
+++ b/tests/clients/test_urllib3.py
@@ -992,3 +992,41 @@ def test_collect_kvs_exception(
         caplog.set_level(logging.DEBUG, logger="instana")
         collect_kvs({}, (), {})
         assert "urllib3 _collect_kvs error: " in caplog.messages
+
+    def test_internal_span_creation_with_url_in_hostname(self) -> None:
+        internal_url = "https://com.instana.example.com/api/test"
+
+        with tracer.start_as_current_span("test"):
+            try:
+                self.http.request("GET", internal_url, retries=False, timeout=1)
+            except Exception:
+                pass
+
+        spans = self.recorder.queued_spans()
+
+        assert len(spans) == 1
+
+        test_span = spans[0]
+        assert test_span.data["sdk"]["name"] == "test"
+
+        urllib3_spans = [span for span in spans if span.n == "urllib3"]
+        assert len(urllib3_spans) == 0
+
+    def test_internal_span_creation_with_url_in_path(self) -> None:
+        internal_url_path = "https://example.com/com.instana/api/test"
+
+        with tracer.start_as_current_span("test"):
+            try:
+                self.http.request("GET", internal_url_path, retries=False, timeout=1)
+            except Exception:
+                pass
+
+        spans = self.recorder.queued_spans()
+
+        assert len(spans) == 1
+
+        test_span = spans[0]
+        assert test_span.data["sdk"]["name"] == "test"
+
+        urllib3_spans = [span for span in spans if span.n == "urllib3"]
+        assert len(urllib3_spans) == 0