database protocol_error ai_generated true

redis.exceptions.ResponseError: MOVED 12345 192.168.1.10:6379

ID: database/redis-cluster-moved-redirect

Also available as: JSON · Markdown · 中文
90%Fix Rate
85%Confidence
1Evidence
2024-01-18First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
Redis 6.2 active — — —
Redis 7.0 active — — —
Redis 7.2 active — — —

Root Cause

A Redis cluster client sent a command to the wrong node; the cluster slot is owned by a different node, requiring a redirect.

generic

中文

Redis集群客户端将命令发送到了错误的节点；该集群槽位由另一个节点拥有，需要重定向。

Official Documentation

https://redis.io/docs/latest/operate/oss_and_stack/management/scaling/

Workarounds

  1. 95% success Use a Redis cluster-aware client (e.g., redis-py-cluster) that automatically follows MOVED redirects: from rediscluster import RedisCluster; rc = RedisCluster(startup_nodes=[{'host':'192.168.1.10','port':6379}], decode_responses=True) 
    Use a Redis cluster-aware client (e.g., redis-py-cluster) that automatically follows MOVED redirects: from rediscluster import RedisCluster; rc = RedisCluster(startup_nodes=[{'host':'192.168.1.10','port':6379}], decode_responses=True)
  2. 90% success Execute the command on the node specified in the MOVED error: redis-cli -h 192.168.1.10 -p 6379 GET mykey 
    Execute the command on the node specified in the MOVED error: redis-cli -h 192.168.1.10 -p 6379 GET mykey
  3. 85% success Refresh the cluster slots map in the client: rc.cluster_slots = rc.cluster_slots() to force a re-fetch of slot assignments. 
    Refresh the cluster slots map in the client: rc.cluster_slots = rc.cluster_slots() to force a re-fetch of slot assignments.

中文步骤

  1. Use a Redis cluster-aware client (e.g., redis-py-cluster) that automatically follows MOVED redirects: from rediscluster import RedisCluster; rc = RedisCluster(startup_nodes=[{'host':'192.168.1.10','port':6379}], decode_responses=True)
  2. Execute the command on the node specified in the MOVED error: redis-cli -h 192.168.1.10 -p 6379 GET mykey
  3. Refresh the cluster slots map in the client: rc.cluster_slots = rc.cluster_slots() to force a re-fetch of slot assignments.

Dead Ends

Common approaches that don't work:

  1. Ignore the MOVED error and retry the same node 100% fail

    The cluster topology is static for that slot; retrying the same node will always fail because the slot is not hosted there.

  2. Manually update the client's cluster slots map by hardcoding IP addresses 70% fail

    Cluster nodes can change IPs or rebalance slots; hardcoding breaks when the cluster topology changes.