The Fix
pip install urllib3==1.26.15
Based on closed urllib3/urllib3 issue #2171 · PR/commit linked
Production note: This usually shows up under retries/timeouts. Treat it as a side-effect risk until you can verify behavior with a canary + real traffic.
@@ -9,6 +9,7 @@
import concurrent.futures
import contextlib
+import errno
import logging
import os
==================================== ERRORS ====================================
______________ ERROR at setup of TestRetryAfter.test_retry_after _______________
Traceback (most recent call last):
File "/home/runner/work/urllib3/urllib3/dummyserver/testcase.py", line 141, in setup_class
cls._start_server()
File "/home/runner/work/urllib3/urllib3/dummyserver/testcase.py", line 129, in _start_server
app, cls.io_loop, cls.certs, cls.scheme, cls.host
File "/home/runner/work/urllib3/urllib3/dummyserver/server.py", line 182, in run_tornado_app
sockets = tornado.netutil.bind_sockets(None, address=host)
File "/home/runner/work/urllib3/urllib3/.nox/test-3-7/lib/python3.7/site-packages/tornado/netutil.py", line 161, in bind_sockets
sock.bind(sockaddr)
OSError: [Errno 98] Address already in use
Re-run the minimal reproduction on your broken version, then apply the fix and re-run.
Option A — Upgrade to fixed release\npip install urllib3==1.26.15\nWhen NOT to use: This fix should not be used in environments where socket binding behavior is critical and cannot tolerate retries.\n\n
Why This Fix Works in Production
- Trigger: ==================================== ERRORS ====================================
- Mechanism: The bind_sockets call fails due to the requested address being already in use in crowded CI environments
- Why the fix works: Adds a retry loop to the bind_sockets call to handle the EADDRINUSE error, which occurs when the requested address is already in use. (first fixed release: 1.26.15).
- If left unfixed, retry loops can amplify load and turn a small outage into a cascade (thundering herd).
Why This Breaks in Prod
- Shows up under Python 3.7 in real deployments (not just unit tests).
- The bind_sockets call fails due to the requested address being already in use in crowded CI environments
- Surfaces as: ==================================== ERRORS ====================================
Proof / Evidence
- GitHub issue: #2171
- Fix PR: https://github.com/urllib3/urllib3/pull/2884
- First fixed release: 1.26.15
- Reproduced locally: No (not executed)
- Last verified: 2026-02-09
- Confidence: 0.95
- Did this fix it?: Yes (upstream fix exists)
- Own content ratio: 0.46
Discussion
High-signal excerpts from the issue thread (symptoms, repros, edge-cases).
“I'm ++ on a retry loop for bind_sockets, easy fix that will likely work too.”
Failure Signature (Search String)
- ==================================== ERRORS ====================================
Error Message
Stack trace
Error Message
-------------
==================================== ERRORS ====================================
______________ ERROR at setup of TestRetryAfter.test_retry_after _______________
Traceback (most recent call last):
File "/home/runner/work/urllib3/urllib3/dummyserver/testcase.py", line 141, in setup_class
cls._start_server()
File "/home/runner/work/urllib3/urllib3/dummyserver/testcase.py", line 129, in _start_server
app, cls.io_loop, cls.certs, cls.scheme, cls.host
File "/home/runner/work/urllib3/urllib3/dummyserver/server.py", line 182, in run_tornado_app
sockets = tornado.netutil.bind_sockets(None, address=host)
File "/home/runner/work/urllib3/urllib3/.nox/test-3-7/lib/python3.7/site-packages/tornado/netutil.py", line 161, in bind_sockets
sock.bind(sockaddr)
OSError: [Errno 98] Address already in use
Minimal Reproduction
==================================== ERRORS ====================================
______________ ERROR at setup of TestRetryAfter.test_retry_after _______________
Traceback (most recent call last):
File "/home/runner/work/urllib3/urllib3/dummyserver/testcase.py", line 141, in setup_class
cls._start_server()
File "/home/runner/work/urllib3/urllib3/dummyserver/testcase.py", line 129, in _start_server
app, cls.io_loop, cls.certs, cls.scheme, cls.host
File "/home/runner/work/urllib3/urllib3/dummyserver/server.py", line 182, in run_tornado_app
sockets = tornado.netutil.bind_sockets(None, address=host)
File "/home/runner/work/urllib3/urllib3/.nox/test-3-7/lib/python3.7/site-packages/tornado/netutil.py", line 161, in bind_sockets
sock.bind(sockaddr)
OSError: [Errno 98] Address already in use
Environment
- Python: 3.7
What Broke
GitHub Actions fails to start tests due to socket binding errors, causing CI job failures.
Why It Broke
The bind_sockets call fails due to the requested address being already in use in crowded CI environments
Fix Options (Details)
Option A — Upgrade to fixed release Safe default (recommended)
pip install urllib3==1.26.15
Use when you can deploy the upstream fix. It is usually lower-risk than long-lived workarounds.
Fix reference: https://github.com/urllib3/urllib3/pull/2884
First fixed release: 1.26.15
Last verified: 2026-02-09. Validate in your environment.
When NOT to Use This Fix
- This fix should not be used in environments where socket binding behavior is critical and cannot tolerate retries.
Verify Fix
Re-run the minimal reproduction on your broken version, then apply the fix and re-run.
Did This Fix Work in Your Case?
Quick signal helps us prioritize which fixes to verify and improve.
Prevention
- Add a CI check that diffs key outputs after upgrades (OpenAPI schema snapshots, JSON payload shapes, CLI output).
- Upgrade behind a canary and run integration tests against the canary before 100% rollout.
- Make timeouts explicit and test them (unit + integration) to avoid silent behavior changes.
- Instrument retries (attempt count + reason) and alert on spikes to catch dependency slowdowns.
Version Compatibility Table
| Version | Status |
|---|---|
| 1.26.15 | Fixed |
Related Issues
No related fixes found.
Sources
We don’t republish the full GitHub discussion text. Use the links above for context.