redis-py Sub-unsub-resub causes redispy to forget the

The Fix

pip install redis==4.0.0

Based on closed redis/redis-py issue #764 · PR/commit linked

Production note: This tends to surface only under concurrency. Reproduce with load tests and watch for lock contention/cancellation paths.

Jump to Verify Open PR/Commit

@@ -2903,7 +2903,9 @@ def reset(self):
             self.connection = None
         self.channels = {}
+        self.pending_unsubscribe_channels = set()
         self.patterns = {}
+        self.pending_unsubscribe_patterns = set()

repro.py

def test_sub_unsub_sub():
    # redis since 2.10.0 to newest (3.0.1)
    import redis

    cli = redis.Redis()
    pubsub = cli.pubsub()
    pubsub.subscribe('test-chan')
    pubsub.unsubscribe('test-chan')
    pubsub.subscribe('test-chan')
    assert pubsub.subscribed

    message = pubsub.get_message()
    assert message
    assert message['channel'] == 'test-chan', message
    assert message['type'] == 'subscribe', message

    message = pubsub.get_message()
    assert message
    assert message['channel'] == 'test-chan', message
    assert message['type'] == 'unsubscribe', message

    message = pubsub.get_message()
    assert message
    assert message['channel'] == 'test-chan', message
    assert message['type'] == 'subscribe', message

    assert pubsub.subscribed  #  <-- this fails!

if __name__ == '__main__':
    test_sub_unsub_sub()

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\npip install redis==4.0.0\nWhen NOT to use: This fix is not suitable if the application requires strict message delivery guarantees.\n\n

Why This Fix Works in Production

Trigger: We recently came across some strange behaviour in our web servers, and tracked the problem down to a race condition in redispy, which is triggered when you…
Mechanism: A race condition occurs when unsubscribing and quickly resubscribing to a channel
Why the fix works: Fixes a race condition in redispy that causes it to forget the channel when unsubscribing and quickly resubscribing. (first fixed release: 4.0.0).

Production impact:

If left unfixed, failures can be intermittent under concurrency (hard to reproduce; shows up as sporadic 5xx/timeouts).

Why This Breaks in Prod

A race condition occurs when unsubscribing and quickly resubscribing to a channel
Production symptom (often without a traceback): We recently came across some strange behaviour in our web servers, and tracked the problem down to a race condition in redispy, which is triggered when you unsubscribe and quickly resubscribe to a channel.

Proof / Evidence

GitHub issue: #764
Fix PR: https://github.com/redis/redis-py/pull/1105
First fixed release: 4.0.0
Reproduced locally: No (not executed)
Last verified: 2026-02-08
Confidence: 0.85
Did this fix it?: Yes (upstream fix exists)
Own content ratio: 0.45

Discussion

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

Jump to Sources Open on GitHub

“I wrote a simple test function which shows a problem. Problem is since 2.10.0 version.”

@marc1n · 2018-12-04 · source

“The ugly consequence of this bug is that in case of reconnect, our channel will be not re-subscribed (because self.channels does not have it)!”

@marc1n · 2018-12-04 · source

“Possible fix (non breaking current semantic of unsubscribe) is to add pending_unsubscribe set which keeps channels to be unsubscribed:”

@marc1n · 2018-12-05 · source

“Is there any contributor, here? I could prepare a pull request, but I see a lot of pull requests (78) nobody cares of - it…”

@marc1n · 2018-12-17 · source

Failure Signature (Search String)

We recently came across some strange behaviour in our web servers, and tracked the problem down to a race condition in redispy, which is triggered when you unsubscribe and quickly
pubsub.subscribe() synchronously adds the channel to pubsub.channels, however pubsub.unsubscribe() doesn't: it waits for the redis server to asynchronously send an unsubscribe

Copy-friendly signature

signature.txt

Failure Signature
-----------------
We recently came across some strange behaviour in our web servers, and tracked the problem down to a race condition in redispy, which is triggered when you unsubscribe and quickly resubscribe to a channel.
pubsub.subscribe() synchronously adds the channel to pubsub.channels, however pubsub.unsubscribe() doesn't: it waits for the redis server to asynchronously send an unsubscribe message.

Error Message

Signature-only (no traceback captured)

error.txt

Error Message
-------------
We recently came across some strange behaviour in our web servers, and tracked the problem down to a race condition in redispy, which is triggered when you unsubscribe and quickly resubscribe to a channel.
pubsub.subscribe() synchronously adds the channel to pubsub.channels, however pubsub.unsubscribe() doesn't: it waits for the redis server to asynchronously send an unsubscribe message.

Minimal Reproduction

repro.py

def test_sub_unsub_sub():
    # redis since 2.10.0 to newest (3.0.1)
    import redis

    cli = redis.Redis()
    pubsub = cli.pubsub()
    pubsub.subscribe('test-chan')
    pubsub.unsubscribe('test-chan')
    pubsub.subscribe('test-chan')
    assert pubsub.subscribed

    message = pubsub.get_message()
    assert message
    assert message['channel'] == 'test-chan', message
    assert message['type'] == 'subscribe', message

    message = pubsub.get_message()
    assert message
    assert message['channel'] == 'test-chan', message
    assert message['type'] == 'unsubscribe', message

    message = pubsub.get_message()
    assert message
    assert message['channel'] == 'test-chan', message
    assert message['type'] == 'subscribe', message

    assert pubsub.subscribed  #  <-- this fails!

if __name__ == '__main__':
    test_sub_unsub_sub()

What Broke

Channels are forgotten, leading to missed messages and unregistered callbacks.

Why It Broke

A race condition occurs when unsubscribing and quickly resubscribing to a channel

Fix Options (Details)

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

pip install redis==4.0.0

When NOT to use: This fix is not suitable if the application requires strict message delivery guarantees.

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

Fix reference: https://github.com/redis/redis-py/pull/1105

First fixed release: 4.0.0

Last verified: 2026-02-08. Validate in your environment.

When NOT to Use This Fix

This fix is not suitable if the application requires strict message delivery guarantees.

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

Add a stress test that runs high-concurrency workloads and fails on thread dumps / blocked locks.
Enable watchdog dumps in prod (faulthandler, thread dump endpoint) to capture deadlocks quickly.

Version Compatibility Table

Version	Status
4.0.0	Fixed

Related Issues

No related fixes found.

Cluster: redis-py:race-condition redis-py hub redis-py best practices All hubs All clusters

Related clusters: Configuration error Retry storm Data consistency

Sources

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

redis-py Sub-unsub-resub causes redispy to forget the channel (Fix)