Redis Cluster

If your infrastructure contains a Redis Cluster, coredis provides a coredis.RedisCluster client than can be used with the same API as coredis.Redis but with awareness of distributing the operations to the appropriate shards.

For operations that operate on single keys the client simply routes the command to the appropriate node.

There are, however a few other categories of operations that cannot simply be used with redis cluster by just routing it to the appropriate node. The API docs will call out any exceptional handling that coredis performs to allow application developers to use the APIs transparently when possible.

Examples of such APIs are:

• async RedisCluster.keys(pattern: StringT = '*') → set[AnyStr]

  Find all keys matching the given pattern

  Returns:

  keys matching pattern.

  Redis command documentation: KEYS

  Cluster note

  The command will be run on all primaries and return the union of the results

• async RedisCluster.exists(keys: Parameters[KeyT]) → int

  Determine if a key exists

  Returns:

  the number of keys that exist from those specified as arguments.

  Redis command documentation: EXISTS

  Cluster note

  The command will be run on all primaries by distributing the keys to the appropriate nodes and return the sum of results.

  To disable this behavior set RedisCluster.non_atomic_cross_slot to False

• async RedisCluster.delete(keys: Parameters[KeyT]) → int

  Delete one or more keys specified by keys

  Returns:

  The number of keys that were removed.

  Redis command documentation: DEL

  Cluster note

  The command will be run on all primaries by distributing the keys to the appropriate nodes and return the sum of results.

  To disable this behavior set RedisCluster.non_atomic_cross_slot to False

• async RedisCluster.script_load(script: StringT) → AnyStr

  Loads a Lua script into the script cache.

  Returns:

  The SHA1 digest of the script added into the script cache

  Redis command documentation: SCRIPT LOAD

  Cluster note

  The command will be run on all nodes and return the response from any shard if all responses are consistent

Replication

coredis supports ensuring synchronous replication of writes using the WAIT command. This is abstracted away with the ensure_replication() context manager.

The following example will ensure that the SET is replicated to atleast 2 replicas within 100 milliseconds (default value of timeout_ms), else raise a ReplicationError:

import asyncio
import coredis


async def test():
    client = coredis.RedisCluster("localhost", 7000)
    with client.ensure_replication(replicas=2):
        await client.set("fubar", 1)

asyncio.run(test())