How to fix TRYAGAIN Multiple keys request during rehashing in Redis

The TRYAGAIN error only appears in Redis Cluster mode during active slot migration. It indicates a temporary condition where your command cannot be executed because the involved keys are split between nodes during resharding.

Verify you are using Redis Cluster:

redis-cli CLUSTER INFO

If the output shows cluster_state:ok, your cluster is healthy. If it shows cluster_state:fail or ongoing migration, a reshard is in progress.

Add exponential backoff retry logic when you encounter TRYAGAIN errors. Most Redis clients provide built-in retry mechanisms.

Node.js (node-redis):

const { createClient } = require('redis');
const client = createClient({
  socket: {
    host: 'localhost',
    port: 6379,
  },
  // Enable automatic retry
  retryStrategy: (retries) => {
    if (retries > 10) return new Error('Max retries exceeded');
    return Math.min(retries * 50, 500); // Exponential backoff
  },
});
await client.connect();

Python (redis-py):

import redis
from redis.retry import Retry
from redis.backoff import ExponentialBackoff

retry = Retry(ExponentialBackoff(), 10)
client = redis.Redis(
  host='localhost',
  port=6379,
  retry=retry,
  retry_on_timeout=True
)

The best way to avoid TRYAGAIN errors is to ensure all keys in a multi-key operation hash to the same slot using Redis hash tags.

Hash tags force keys to the same slot by extracting only the text between curly braces for hash computation:

# Without hash tags - keys may be on different slots
redis.mset({'user:1:name': 'Alice', 'user:2:name': 'Bob'})

# With hash tags - keys map to the same slot
redis.mset({'{user:1}:name': 'Alice', '{user:1}:profile': '{...}'})
redis.mset({'{user:2}:name': 'Bob', '{user:2}:settings': '{...}'})

The {user:1} portion (inside braces) is hashed, so all keys with the same tag are guaranteed on the same slot.

If you control the cluster, you can monitor resharding progress:

# Check cluster status
redis-cli CLUSTER NODES

# Monitor migrations in progress
redis-cli CLUSTER SLOTS

Once all slots show OK state and migrations complete, TRYAGAIN errors will cease. If you initiated a reshard manually via redis-cli CLUSTER MEET or a cluster manager, wait for the operation to finish before running multi-key commands on keys spanning different slots.

If you frequently encounter TRYAGAIN errors during normal cluster operations, consider your resharding approach:

Option 1: Disable auto-resharding during peak hours
Postpone manual cluster rebalancing to maintenance windows when traffic is low.

Option 2: Use gradual migration
Some cluster managers allow slow/graceful migrations that minimize disruptions.

Option 3: Increase timeout values
Allow clients more time for retry-able operations:

import redis
client = redis.Redis(
  host='localhost',
  port=6379,
  socket_timeout=10,  # Longer timeout for client operations
  socket_keepalive=True,
)