[EventHub] Fix aiohttp ClientSession leak in WebSocketTransportAsync close/connect

[MuhammadAliShahzad]

Summary

Fixes #46828.

WebSocketTransportAsync.close() calls self.sock.close() and self.session.close() sequentially without try/except. When sock.close() raises (e.g. the websocket is already disconnected or timed out), session.close() never runs and the aiohttp.ClientSession leaks. The sibling AsyncTransport.close() in the same file already handles this correctly — this PR brings WebSocketTransportAsync in line with that pattern.

WebSocketTransportAsync.connect() had two related leaks:

• On reconnect, a previously-assigned self.session was not closed before being overwritten.
• If ws_connect() raised anything other than ClientConnectorError, the newly-created session was not cleaned up.

Changes

• azure/eventhub/_pyamqp/aio/_transport_async.py
  • close(): wrap sock.close() and session.close() independently in try/except, log via _LOGGER.debug(..., extra=self.network_trace_params) (same pattern as AsyncTransport.close() at line 370–385), and null state after close. self.connected = False moved outside the async with block to match the sibling.
  • connect(): close any previous self.session before assigning a new one; add a broad except Exception: clause that cleans up the new session before re-raising; refactor session cleanup into a small _close_session_safely() helper used by all three call sites.
• tests/pyamqp_tests/asynctests/test_websocket_transport_async_unit.py (new)
  • 6 unit tests that mock aiohttp.ClientSession and assert the cleanup invariants — no live Event Hubs needed.
• CHANGELOG.md: bullet under a new 5.15.2 (Unreleased) ### Bugs Fixed section.

Why the sibling sync transport is not affected

The sync WebSocketTransport in _transport.py uses the websocket-client library, not aiohttp, and does not maintain a separate ClientSession. Its close() (line 743) and connect() (line 655) already perform proper cleanup. So this fix is async-only.

Test plan

• New unit tests fail against the previous (unfixed) code, for the right reasons (mocked sock.close() raises → session.close() never called)
• All 6 new unit tests pass against the fixed code
• mypy reports zero new errors in the modified file
• pyright reports zero new errors in the modified file (pre-existing reportOptionalMemberAccess in unchanged _read/_write remain)
• pylint clean for the modified file (only unrelated unknown-option-value warnings from the pylint-guidelines-checker plugin in my local env)
• CI (azpysdk checks) — to be validated on PR run

Reproduction of the original bug

Run an EventHub consumer/producer over WebSocket transport (TransportType.AmqpOverWebsocket) for an extended period with intermittent network disruptions. The Unclosed client session errors appear in logs on each reconnection cycle and on shutdown.

🤖 Generated with Claude Code

[github-actions]

Thank you for your contribution @MuhammadAliShahzad! We will review the pull request and get back to you soon.

[Copilot]

Pull request overview

Fixes aiohttp.ClientSession resource leaks in the async WebSocket transport for azure-eventhub by making close()/connect() exception-safe and adding unit coverage to prevent regressions.

Changes:

• Made WebSocketTransportAsync.connect() close any prior session before reconnect and ensure new sessions are closed on unexpected ws_connect() failures.
• Made WebSocketTransportAsync.close() resilient to exceptions from websocket/session close so cleanup still completes and state is reset.
• Added focused unit tests for cleanup invariants and documented the fix in the changelog.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
sdk/eventhub/azure-eventhub/azure/eventhub/_pyamqp/aio/_transport_async.py	Adds safe session cleanup helper; hardens connect()/close() against exceptions to prevent ClientSession leaks.
sdk/eventhub/azure-eventhub/tests/pyamqp_tests/asynctests/test_websocket_transport_async_unit.py	New unit tests validating session/socket cleanup behavior across failure paths.
sdk/eventhub/azure-eventhub/CHANGELOG.md	Notes the leak fix under a new 5.15.2 (Unreleased) entry.

[MuhammadAliShahzad]

Thanks for the review. Applied the suggestion in 5394f07 — close() now calls _close_session_safely() instead of duplicating the session-close try/except. All 6 unit tests still pass.

[MuhammadAliShahzad]

@microsoft-github-policy-service agree

[MuhammadAliShahzad]

Pushed 7e8fbc3 to address Build Analyze failures:

• Verify ChangeLogEntries: bumped VERSION in _version.py from 5.15.1 → 5.15.2 (so the verifier matches my new top-of-file entry) and trimmed the empty ### Features Added / ### Breaking Changes / ### Other Changes subsections.
• Run MyPy: the failures are all pre-existing on main in files not touched by this PR (_common.py, _producer.py, _consumer.py, _consumer_async.py). Verified by running the same mypy command locally against upstream/main. Happy to defer to maintainers on whether those should be fixed in a separate PR or whether the CI mypy check needs to be relaxed.

[j7nw4r]

Mind rebasing on current main? _common.py (one of your "pre-existing mypy" files) got a commit since your merge base, so a rebase will either clear Build Analyze or validate the claim against today's main. No conflicts; _transport_async.py itself hasn't moved.

[MuhammadAliShahzad]

Rebased onto current main. Now that #46988 is merged, the previously-failing mypy errors in _common.py (and the other untouched files) are fixed on main, so Build Analyze should come back green on this run. _transport_async.py itself was unchanged by the rebase.
@j7nw4r