uvicorn Development using gunicorn + uvicorn (Fix)

The Fix

Upgrade to version 0.11.4 or later.

Based on closed Kludex/uvicorn issue #188 · PR/commit linked

Jump to Verify Open PR/Commit

@@ -26,8 +26,9 @@ def isclosed(self):
 
 class _ASGIAdapter(requests.adapters.HTTPAdapter):
-    def __init__(self, app: typing.Callable) -> None:
+    def __init__(self, app: typing.Callable, raise_server_exceptions=True) -> None:
         self.app = app

repro.py

import asyncio
import logging
# import warnings


LOGGER = logging.getLogger()


async def never_called():
    LOGGER.debug("I'm never actually called.")


async def uncomsumed():
    LOGGER.debug("My exception is never consumed.")
    raise ValueError("Ouch")


async def transport():
    LOGGER.debug("Rapid")
    connect = asyncio.open_connection("http://google.com", 80)
    reader, writer = await connect


async def create():
    await asyncio.sleep(3.0)
    LOGGER.debug("(1) create file")


async def write():
    await asyncio.sleep(1.0)
    LOGGER.debug("(2) write into file")


async def close():
    LOGGER.debug("(3) close file")


async def pending():
    asyncio.ensure_future(create())
    asyncio.ensure_future(write())
    asyncio.ensure_future(close())
    await asyncio.sleep(2.0)
    loop.stop()


if __name__ == "__main__":
    # warnings.simplefilter("always", ResourceWarning)
    logging.basicConfig(level="DEBUG",
                        format="[%(name)s] [%(levelname)s] %(message)s")
    loop = asyncio.get_event_loop()
    # loop.set_debug(True)
    never_called()
    asyncio.ensure_future(uncomsumed())
    asyncio.ensure_future(transport())
    asyncio.ensure_future(pending())
    loop.run_forever()
    LOGGER.warning("Pending tasks at exit: %r.", asyncio.Task.all_tasks(loop))
    loop.close()

verify

Re-run the minimal reproduction on your broken version, then apply the fix and re-run.

fix.md

Option A — Upgrade to fixed release\nUpgrade to version 0.11.4 or later.\nWhen NOT to use: This fix is not suitable if you need to suppress exceptions intentionally.\n\n

Why This Fix Works in Production

Trigger: I'm able to get tracebacks about uncalled co-routines via environment variable `PYTHONASYNCIODEBUG=1` or inside the application with:
Mechanism: DebugMiddleware was swallowing exceptions when returning 500 responses
Why the fix works: Resolves an issue where DebugMiddleware was swallowing exceptions when returning 500 responses, allowing them to be logged properly. (first fixed release: 0.11.4).

Why This Breaks in Prod

Shows up under Python 3.6 in real deployments (not just unit tests).
DebugMiddleware was swallowing exceptions when returning 500 responses
Production symptom (often without a traceback): I'm able to get tracebacks about uncalled co-routines via environment variable `PYTHONASYNCIODEBUG=1` or inside the application with:

Proof / Evidence

GitHub issue: #188
Fix PR: https://github.com/encode/uvicorn/pull/190
First fixed release: 0.11.4
Reproduced locally: No (not executed)
Last verified: 2026-02-09
Confidence: 0.85
Did this fix it?: Yes (upstream fix exists)
Own content ratio: 0.39

Discussion

High-signal excerpts from the issue thread (symptoms, repros, edge-cases).

Jump to Sources Open on GitHub

“Sorry, I’m slightly lost”

@lovelydinosaur · 2018-09-05 · confirmation · source

“> I guess my only remaining problem is that I'm raising an exception (ValueError) in one of my routes and I had expected to see…”

@lovelydinosaur · 2018-09-06 · confirmation · source

“> Sorry, I’m slightly lost”

@Midnighter · 2018-09-05 · confirmation · source

“Probably not especially related but worth noting - uvicorn uses the root logger in 0.3.2, but uses a properly named logger in 0.3.3.”

@lovelydinosaur · 2018-09-05 · source

Failure Signature (Search String)

I'm able to get tracebacks about uncalled co-routines via environment variable `PYTHONASYNCIODEBUG=1` or inside the application with:
Adding the handler on the `uvicorn` logger and separately changing the level should not be necessary due to the changes on the `root` logger but I was kinda desperate to try

Copy-friendly signature

signature.txt

Failure Signature
-----------------
I'm able to get tracebacks about uncalled co-routines via environment variable `PYTHONASYNCIODEBUG=1` or inside the application with:
Adding the handler on the `uvicorn` logger and separately changing the level should not be necessary due to the changes on the `root` logger but I was kinda desperate to try something.

Error Message

Signature-only (no traceback captured)

error.txt

Error Message
-------------
I'm able to get tracebacks about uncalled co-routines via environment variable `PYTHONASYNCIODEBUG=1` or inside the application with:
Adding the handler on the `uvicorn` logger and separately changing the level should not be necessary due to the changes on the `root` logger but I was kinda desperate to try something.

Minimal Reproduction

repro.py

import asyncio
import logging
# import warnings


LOGGER = logging.getLogger()


async def never_called():
    LOGGER.debug("I'm never actually called.")


async def uncomsumed():
    LOGGER.debug("My exception is never consumed.")
    raise ValueError("Ouch")


async def transport():
    LOGGER.debug("Rapid")
    connect = asyncio.open_connection("http://google.com", 80)
    reader, writer = await connect


async def create():
    await asyncio.sleep(3.0)
    LOGGER.debug("(1) create file")


async def write():
    await asyncio.sleep(1.0)
    LOGGER.debug("(2) write into file")


async def close():
    LOGGER.debug("(3) close file")


async def pending():
    asyncio.ensure_future(create())
    asyncio.ensure_future(write())
    asyncio.ensure_future(close())
    await asyncio.sleep(2.0)
    loop.stop()


if __name__ == "__main__":
    # warnings.simplefilter("always", ResourceWarning)
    logging.basicConfig(level="DEBUG",
                        format="[%(name)s] [%(levelname)s] %(message)s")
    loop = asyncio.get_event_loop()
    # loop.set_debug(True)
    never_called()
    asyncio.ensure_future(uncomsumed())
    asyncio.ensure_future(transport())
    asyncio.ensure_future(pending())
    loop.run_forever()
    LOGGER.warning("Pending tasks at exit: %r.", asyncio.Task.all_tasks(loop))
    loop.close()

Environment

Python: 3.6

What Broke

Users were not seeing exceptions in logs during error responses.

Why It Broke

DebugMiddleware was swallowing exceptions when returning 500 responses

Fix Options (Details)

Option A — Upgrade to fixed release Safe default (recommended)

Upgrade to version 0.11.4 or later.

When NOT to use: This fix is not suitable if you need to suppress exceptions intentionally.

Use when you can deploy the upstream fix. It is usually lower-risk than long-lived workarounds.

Fix reference: https://github.com/encode/uvicorn/pull/190

First fixed release: 0.11.4

Last verified: 2026-02-09. Validate in your environment.

When NOT to Use This Fix

This fix is not suitable if you need to suppress exceptions intentionally.

Verify Fix

verify

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

Capture the exact failing error string in logs and tests so you can reproduce via a minimal script.
Pin production dependencies and upgrade only with a reproducible test that hits the failing path.

Version Compatibility Table

Version	Status
0.11.4	Fixed

Related Issues

No related fixes found.

Cluster: uvicorn:other All hubs All clusters

Related clusters: Configuration error Data consistency Race condition

Sources

We don’t republish the full GitHub discussion text. Use the links above for context.