AttributeError during Redis Cluster create-before-destroy process

[QuentinBrosse]

Summary

During a GCP Memorystore for Redis Cluster maintenance, nodes were replaced as part of a create-before-destroy process. During this transition, the client’s in-memory cluster topology became temporarily inconsistent with the actual cluster state, resulting in _determine_nodes() returning a None value for a target node. It resulted in AttributeError: 'NoneType' object has no attribute 'name' originating from this line.

We reproduced this issue by scaling down our Redis cluster.

Environment

• redis-py version: 5.3.1
• Python version: 3.11
• Deployment environment: Google Cloud Memorystore for Redis Cluster
• Topology: multi-node Redis Cluster
• Client type: redis.asyncio.RedisCluster

Full stack trace

builtins.AttributeError: 'NoneType' object has no attribute 'name'

Traceback (most recent call last):
  File "/usr/local/lib/python3.11/site-packages/ddtrace/_trace/utils_redis.py", line 96, in _instrument_redis_execute_async_cluster_pipeline
    yield span
  File "/usr/local/lib/python3.11/site-packages/ddtrace/contrib/internal/redis/asyncio_patch.py", line 36, in instrumented_async_execute_cluster_pipeline
    return await func(*args, **kwargs)
  File "/usr/local/lib/python3.11/site-packages/redis/asyncio/cluster.py", line 1488, in execute
  File "/usr/local/lib/python3.11/site-packages/redis/asyncio/cluster.py", line 1571, in _execute
AttributeError: 'NoneType' object has no attribute 'name'

Reproduction steps

1.	Start a Redis Cluster with multiple shards.
2.	Connect using `redis.asyncio.RedisCluster`.
3.	Continuously run pipelined commands.
4.	Scale up/down cluster
5.	Observe intermittent `AttributeError` exceptions from the client.

Expected behavior

redis-py should handle temporary None nodes gracefully — either by:
• retrying the command after topology refresh, or
• raising a more specific error we can use to retry.

Questions

• I see the _determine_nodes() method changed a lot in the newer versions. Could upgrading the version could fix that?
• Is it a known issue or something we should fix in this lib?

Thanks!

[petyaslavova]

Hi @QuentinBrosse, thank for reporting this! I will try to answer your questions by the end of the week.