Crash with RedisCluster and RedisDict's update

[mlasevich]

When using RedisDict with an instance of redis.cluster.RedisCluster() - a crash occurs in base.py:

 Fraceback (most recent call last):
  File "/app/poc3.py", line 28, in <module>
    demo()
  File "/app/poc3.py", line 24, in demo
    data.update({'key': 'value'})
  File "/usr/local/lib/python3.9/site-packages/pottery/dict.py", line 144, in update
    with self._watch(arg) as pipeline:
  File "/usr/local/lib/python3.9/contextlib.py", line 119, in __enter__
    return next(self.gen)
  File "/usr/local/lib/python3.9/site-packages/pottery/base.py", line 232, in _watch
    for context_manager in self.__context_managers(*others):
  File "/usr/local/lib/python3.9/site-packages/pottery/base.py", line 220, in __context_managers
    redises[connection_args(container.redis)].append(container)
  File "/usr/local/lib/python3.9/site-packages/pottery/base.py", line 89, in connection_args
    redis.connection_pool.connection_kwargs['host'],
AttributeError: 'RedisCluster' object has no attribute 'connection_pool'

To Reproduce

here is simple code to reproduce:

import os

from pottery import RedisDict

from redis.cluster import RedisCluster

env = os.environ

redis = RedisCluster(
    host=env.get('REDIS_HOST'),
    port=int(env.get('REDIS_PORT')),
    username=env.get('REDIS_USER'),
    password=env.get('REDIS_PASSWORD'),
    charset='utf-8',
    ssl_cert_reqs='none',
    ssl=True,
    skip_full_coverage_check=True
)

def demo():
    data = RedisDict(key=f"data:test", redis=redis)
    data.update({'key': 'value'})

if __name__ == "__main__":
    demo()

Expected behavior

I expect it to work same regardless if it is a cluster or a single Redis instance :-/

Environment (please complete the following information):

• OS: Linux
• Python version 3.9.18
• Redis version : AWS MemoryDB 6.x cluster with a single node