fix(client_reqrep): clear _traces in close/release to break session reference cycle

[prshant70]

What

Clears self._traces = [] in ClientResponse.close() and ClientResponse.release() to deterministically break a circular reference that causes slow memory growth (issue #11043).

A CHANGES/11043.bugfix towncrier fragment is also included.

Closes #11043

Why

Issue #11043 reports slow, steady RSS growth when making repeated HTTP requests with tracing enabled. The root cause is a reference cycle that prevents CPython's reference-counting from reclaiming ClientResponse objects promptly:

ClientResponse._traces  ->  [Trace, ...]
Trace._session          ->  ClientSession
ClientSession           ->  (keeps other things alive)

_cleanup_writer() already clears self._session = None, but self._traces was never cleared, keeping every Trace (and therefore the session) alive until the cycle GC ran — which is delayed and non-deterministic under load.

How

self._traces = [] is added to close() and release() — not to _cleanup_writer().

The reason _cleanup_writer() is excluded: it is also called from _response_eof(), which fires synchronously as a payload EOF callback during await self.content.read(). At that point read() has not yet iterated _traces to fire on_response_chunk_received hooks. Clearing _traces there silently dropped all response chunk trace callbacks (confirmed by test_request_tracing and test_request_tracing_url_params regressions).

close() and release() are always called after the user finishes consuming the response (via context manager __aexit__), so the cycle is still broken deterministically — just at the right time.

Key files:

• aiohttp/client_reqrep.py — self._traces = [] added to close() and release()
• CHANGES/11043.bugfix — towncrier changelog fragment
• tests/test_client_response.py — four regression tests (including one that asserts _response_eof() intentionally does NOT clear _traces)

Testing

pytest tests/test_client_response.py tests/test_client_session.py -q

New tests:

• test_close_clears_traces_and_session — _traces == [] and _session is None after close()
• test_release_clears_traces_and_session — same after release()
• test_response_eof_clears_session_but_not_traces — _session is None but _traces preserved after _response_eof(), so trace callbacks can still fire
• test_no_circular_ref_after_close — CPython-only weakref test confirming the object is collected after close()

Checklist

• Tests pass (pytest tests/test_client_response.py tests/test_client_session.py -q)
• Changelog entry added (CHANGES/11043.bugfix)
• No new abstractions or unnecessary changes

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.95%. Comparing base (1f005f1) to head (07ec948).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #12669   +/-   ##
=======================================
  Coverage   98.95%   98.95%           
=======================================
  Files         131      131           
  Lines       46688    46731   +43     
  Branches     2421     2421           
=======================================
+ Hits        46200    46243   +43     
  Misses        366      366           
  Partials      122      122           

Flag	Coverage Δ
Autobahn	22.41% <16.27%> (-0.01%)	⬇️
CI-GHA	98.92% <100.00%> (+<0.01%)	⬆️
OS-Linux	98.67% <100.00%> (-0.01%)	⬇️
OS-Windows	97.04% <100.00%> (+<0.01%)	⬆️
OS-macOS	97.93% <100.00%> (-0.01%)	⬇️
Py-3.10	98.16% <100.00%> (+<0.01%)	⬆️
Py-3.11	98.41% <100.00%> (-0.01%)	⬇️
Py-3.12	98.50% <100.00%> (-0.01%)	⬇️
Py-3.13	98.48% <100.00%> (-0.01%)	⬇️
Py-3.14	98.49% <100.00%> (-0.01%)	⬇️
Py-3.14t	97.55% <100.00%> (+<0.01%)	⬆️
Py-pypy-3.11	97.40% <76.74%> (-0.04%)	⬇️
VM-macos	97.93% <100.00%> (-0.01%)	⬇️
VM-ubuntu	98.67% <100.00%> (-0.01%)	⬇️
VM-windows	97.04% <100.00%> (+<0.01%)	⬆️
cython-coverage	37.91% <4.65%> (-0.04%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[codspeed-hq]

Merging this PR will not alter performance

✅ 72 untouched benchmarks
⏩ 72 skipped benchmarks¹

---

_{Comparing prshant70:fix/client-response-circular-ref-cleanup (07ec948) with master (1f005f1)}

Footnotes

1. 72 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩