MCP OAuth and HTTP transport reconnect deadlock fixes
Cluster 31987 · mcp. Data as of
2 open items. Fix merged in #97458 on 29 Aug 2026. Check it covers your case.
The short answer
- Known?
- Yes. 10 related items grouped here, 2 issues already closed.
- Fix event
- A fix was merged to main in PR #97458. Check it covers your case. No release data is tracked.
- Tracker
- The maintainers' tracker marks this
unresolved. That is separate from the fix event. - Still open
- 1 issue and 1 PR.
Next step: read #97458, then check whether your report is the one still open.
Read #97458Pick up a Lab taskThese issues and fixes address deadlocks and reconnect failures in MCP OAuth-backed servers, specifically around stranded auth-flow generator locks and anyio cleanup errors during streamable HTTP transport teardown. Multiple fix attempts and salvage PRs suggest ongoing iteration on the same underlying concurrency problems.
Open issues1
- Issue#31987MCP HTTP transport: anyio RuntimeError during streamable_http_client cleanup causes reconnect failure loop
Open PRs1
- PR#38198fix: Close inner SDK auth generator to prevent OAuth reconnect deadlock
Fixes on record
-
#97458 fix(mcp-oauth): avoid locking across resource requests
files:
tests/tools/test_mcp_oauth_bidirectional.py,tests/tools/test_mcp_oauth_manager.py,tools/mcp_oauth_manager.py
5 PRs closed without merging
From the close comments: superseded by another PR 3 · done elsewhere 1 · stale or abandoned 1
- #32111 fix(tools): ignore anyio cleanup artifact during MCP HTTP reconnect · stale or abandoned
-
#63495 fix(mcp-oauth): close inner SDK auth generator on teardown (salvage #38198) · superseded by another PR →
#97458,#90888 -
#76526 fix: close stranded MCP OAuth auth flows · superseded by another PR →
#97458,#91049 -
#90888 fix(mcp-oauth): bind delegated auth-flow teardown to the owning task · done elsewhere →
7c6f5ce,b2c4f1f,#38193,#84963,#88875 -
#91049 fix(mcp-oauth): recover stranded auth-flow locks · superseded by another PR →
#97458
Closed issues · 2
- Closed#38193[Bug]: OAuth-backed MCP server permanently deadlocks after keepalive reconnect: auth-flow generator's lock is released cross-task.
- Closed#49543OAuth MCP servers (e.g. Honeycomb) drop from active toolset mid-session — RuntimeError: The current task is not holding this lock + 120s call hangs
How this cluster was grouped
-
Link evidence (
partly structural links): some members are connected by structural evidence (shared changed files, or a closing/superseded reference); the rest are joined by an automated relevance judgement over their text. - Members outside that structural core may belong to a different underlying problem.
-
Pairwise link density
0.17: the share of item pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36.1.00means every pair is linked.
Pick up a Lab task
Check whether a fix covers the open issues here, with your own agent. Bring back evidence for maintainers.
Open the LabAt a glance
- Subsystem
- mcp
- Merged PRs
- #97458
- Closed issues
- 2
- Attempts
- 5 PRs closed without merging
- Fix link checked
- 29 Aug 2026