菜单
Master Redis Cluster - horizontal scaling, hash slots, resharding, cluster management, and distributed architecture
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/pluginagentmarketplace/custom-plugin-redis --skill redis-cluster命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | redis-cluster |
| description | Master Redis Cluster - horizontal scaling, hash slots, resharding, cluster management, and distributed architecture |
| sasmp_version | 1.3.0 |
| bonded_agent | 06-redis-clustering |
| bond_type | PRIMARY_BOND |
| version | 2.1.0 |
| last_updated | 2025-01 |
| parameters | {"cluster_operation":{"type":"string","required":true,"enum":["create","add_node","remove_node","reshard","rebalance","failover"]},"node_count":{"type":"integer","required":false,"min":6,"description":"Minimum 6 nodes for production (3 masters + 3 replicas)"},"replicas_per_master":{"type":"integer","required":false,"default":1,"min":1,"max":3}} |
| retry_config | {"max_retries":5,"backoff_strategy":"exponential","backoff_base_ms":2000,"retryable_errors":["CLUSTERDOWN","MOVED","ASK","connection_timeout"]} |
| observability | {"logging":{"level":"info","log_redirects":true},"metrics":["cluster_state","slots_assigned","nodes_ok","replication_lag_ms"]} |
| validation | {"cluster_health":{"all_slots_covered":true,"min_replicas_per_master":1},"operations":{"require_backup_before_reshard":true}} |
Production-grade Redis Cluster management for horizontal scaling. Master hash slot distribution, node management, resharding, and failover procedures.
┌─────────────────────────────────────────────────────────────────────┐
│ REDIS CLUSTER (16384 Hash Slots) │
├─────────────────────────────────────────────────────────────────────┤
│ │
│ ┌──────────────┐ ┌──────────────┐ ┌──────────────┐ │
│ │ Master 1 │ │ Master 2 │ │ Master 3 │ │
│ │ Slots 0-5460 │ │ Slots 5461- │ │ Slots 10923- │ │
│ │ :6379 │ │ 10922 │ │ 16383 │ │
│ └──────┬───────┘ └──────┬───────┘ └──────┬───────┘ │
│ │ │ │ │
│ ▼ ▼ ▼ │
│ ┌──────────────┐ ┌──────────────┐ ┌──────────────┐ │
│ │ Replica 1 │ │ Replica 2 │ │ Replica 3 │ │
│ │ :6380 │ │ :6381 │ │ :6382 │ │
│ └──────────────┘ └──────────────┘ └──────────────┘ │
│ │
│ Key Routing: CRC16(key) % 16384 → Slot → Master │
│ │
└─────────────────────────────────────────────────────────────────────┘
# Each node's redis.conf
port 6379
cluster-enabled yes
cluster-config-file nodes.conf
cluster-node-timeout 5000
appendonly yes
# Create 6-node cluster (3 masters + 3 replicas)
redis-cli --cluster create \
192.168.1.1:6379 192.168.1.2:6379 192.168.1.3:6379 \
192.168.1.4:6379 192.168.1.5:6379 192.168.1.6:6379 \
--cluster-replicas 1
# Confirm with 'yes'
CLUSTER INFO # Cluster state summary
CLUSTER NODES # All nodes list
CLUSTER SLOTS # Slot assignments
CLUSTER MYID # Current node ID
CLUSTER KEYSLOT key # Which slot for key
CLUSTER COUNTKEYSINSLOT slot # Keys in slot
CLUSTER GETKEYSINSLOT slot count # Get keys from slot
redis-cli --cluster add-node new-node:6379 existing-node:6379
redis-cli --cluster add-node new-node:6379 existing-node:6379 \
--cluster-slave --cluster-master-id <master-node-id>
redis-cli --cluster del-node host:6379 <node-id>
redis-cli --cluster check existing-node:6379
redis-cli --cluster fix existing-node:6379
# Interactive reshard
redis-cli --cluster reshard existing-node:6379
# Scripted reshard
redis-cli --cluster reshard existing-node:6379 \
--cluster-from <source-node-id> \
--cluster-to <dest-node-id> \
--cluster-slots 1000 \
--cluster-yes
# Rebalance (auto-distribute slots)
redis-cli --cluster rebalance existing-node:6379 \
--cluster-use-empty-masters
# Force failover (from replica)
redis-cli -c -h replica-host CLUSTER FAILOVER
# Force failover (ignore master state)
redis-cli -c -h replica-host CLUSTER FAILOVER FORCE
# Takeover (don't wait for master agreement)
redis-cli -c -h replica-host CLUSTER FAILOVER TAKEOVER
# Keys with same hash tag go to same slot
SET {user:123}:profile "..."
SET {user:123}:sessions "..."
SET {user:123}:preferences "..."
# Multi-key operations work with hash tags
MGET {user:123}:profile {user:123}:sessions
DEL {user:123}:profile {user:123}:sessions
Pattern Hash Part Slot
user:123:profile user:123 varies
{user:123}:profile user:123 same
user:{123}:profile 123 same
order:{customer:456}:item customer:456 same
from redis.cluster import RedisCluster
# Connect to cluster
rc = RedisCluster(
host='node1',
port=6379,
decode_responses=True
)
# Auto-follows redirects
rc.set('key', 'value')
rc.get('key')
# Pipeline with hash tags
pipe = rc.pipeline()
pipe.set('{user:1}:name', 'Alice')
pipe.set('{user:1}:email', 'alice@example.com')
pipe.execute()
# MOVED - permanent redirect
GET key
# (error) MOVED 5649 192.168.1.2:6379
# ASK - temporary redirect (during migration)
GET key
# (error) ASK 5649 192.168.1.2:6379
# Must send ASKING before retry
# redis.conf for cluster node
# Cluster settings
cluster-enabled yes
cluster-config-file nodes.conf
cluster-node-timeout 5000
cluster-replica-validity-factor 10
cluster-migration-barrier 1
cluster-require-full-coverage yes
cluster-replica-no-failover no
# Network (cluster bus = port + 10000)
port 6379
bind 0.0.0.0
cluster-announce-ip 192.168.1.1
cluster-announce-port 6379
cluster-announce-bus-port 16379
# Persistence
appendonly yes
appendfsync everysec
# Memory
maxmemory 4gb
maxmemory-policy allkeys-lru
cluster-config.conf - Production cluster node configconfig.yaml - Cluster settingscluster-health.sh - Health check scripthelper.py - Python utilitiesCLUSTER_GUIDE.md - Complete setup guideGUIDE.md - Reference documentationCLUSTERDOWN The cluster is down
Diagnosis:
redis-cli -c CLUSTER INFO | grep cluster_state
redis-cli -c CLUSTER NODES | grep -v connected
Causes & Fixes:
| Cause | Fix |
|---|---|
| Slots not covered | Add nodes or assign slots |
| Master down, no replica | Restore node or failover |
| Network partition | Fix network |
| Quorum lost | Restore majority |
# Check uncovered slots
redis-cli --cluster check node:6379
# Fix automatically
redis-cli --cluster fix node:6379 --cluster-fix-with-unreachable-masters
Cause: Stale cluster topology in client
Fix:
# Refresh cluster slots
rc.cluster_nodes()
# Or recreate connection
rc = RedisCluster(host='node1', port=6379)
CROSSSLOT Keys in request don't hash to the same slot
Fix: Use hash tags
# Before (fails)
MGET user:1:name user:2:name
# After (works)
MGET {user:1}:name {user:1}:email
# Check migration status
redis-cli -c -h source-node CLUSTER NODES | grep migrating
redis-cli -c -h dest-node CLUSTER NODES | grep importing
# Force finish
redis-cli -c -h dest-node CLUSTER SETSLOT slot STABLE
redis-cli -c -h source-node CLUSTER SETSLOT slot STABLE
□ All 16384 slots assigned?
□ All masters have replicas?
□ Cluster bus ports open (port+10000)?
□ cluster-announce-ip set correctly?
□ Network allows all node communication?
□ nodes.conf not corrupted?
□ Using cluster-aware client?
□ Hash tags for multi-key ops?
#!/bin/bash
# check-cluster-health.sh
NODE=$1
echo "=== Cluster Health Check ==="
# Check state
echo -n "Cluster state: "
redis-cli -c -h $NODE CLUSTER INFO | grep cluster_state
# Check slots
echo -n "Slots OK: "
redis-cli -c -h $NODE CLUSTER INFO | grep cluster_slots_ok
# Check nodes
echo "Nodes:"
redis-cli -c -h $NODE CLUSTER NODES | awk '{print $1, $2, $3}'
# Run check
echo "Running cluster check..."
redis-cli --cluster check $NODE:6379
| Code | Name | Description | Recovery |
|---|---|---|---|
| C001 | CLUSTERDOWN | Cluster unavailable | Check nodes, fix slots |
| C002 | MOVED | Key on different node | Follow redirect |
| C003 | ASK | Slot being migrated | ASKING + retry |
| C004 | CROSSSLOT | Keys in different slots | Use hash tags |
| C005 | TRYAGAIN | Slot in migration | Retry later |
# test_redis_cluster.py
import redis
from redis.cluster import RedisCluster
import pytest
@pytest.fixture
def rc():
return RedisCluster(host='node1', port=6379, decode_responses=True)
def test_cluster_info(rc):
info = rc.cluster_info()
assert info['cluster_state'] == 'ok'
assert info['cluster_slots_ok'] == 16384
def test_set_get(rc):
rc.set('cluster:test', 'value')
assert rc.get('cluster:test') == 'value'
rc.delete('cluster:test')
def test_hash_tags(rc):
rc.set('{user:1}:name', 'Alice')
rc.set('{user:1}:email', 'alice@example.com')
result = rc.mget('{user:1}:name', '{user:1}:email')
assert result == ['Alice', 'alice@example.com']
rc.delete('{user:1}:name', '{user:1}:email')
def test_pipeline_hash_tags(rc):
pipe = rc.pipeline()
pipe.set('{order:1}:status', 'pending')
pipe.set('{order:1}:items', '["item1"]')
results = pipe.execute()
assert all(results)
rc.delete('{order:1}:status', '{order:1}:items')