Redisson 3.6.1在AWS ElastiCache环境频繁抛出RedisTimeoutException求助

Let's break down the common causes and actionable fixes for the org.redisson.client.RedisTimeoutException you're facing—especially since this happens even under low load with a simple get/put workload on an LRU-bounded RMapCache:

1. Outdated Redisson Version Compatibility Issues #

Redisson 3.6.1 is a fairly old release (circa 2018), and it might not play nicely with newer AWS ElastiCache Redis versions (like 6.x or 7.x). Older client versions often lack support for modern Redis protocol features, or have unpatched bugs in connection handling that lead to timeouts during node interactions or failovers.

Fix: Upgrade Redisson to a recent stable version (e.g., 3.20.x or later). Newer releases include improved compatibility with ElastiCache, smarter failover handling, and fixes for connection pool-related bugs that could trigger timeouts.

2. Suboptimal Connection Pool Configuration #

Redisson's default connection pool settings might not be tuned for AWS's network environment, even with low load. If your application is exhausting connections or waiting too long for idle connections, timeouts will occur.

Recommended Configuration Adjustments (example for master-slave setup):

masterSlaveConfig:
  idleConnectionTimeout: 10000
  connectTimeout: 10000
  timeout: 5000  # Increase command timeout to account for minor network delays
  retryAttempts: 3
  retryInterval: 1500
  connectionMinimumIdleSize: 10  # Maintain enough idle connections for low-load scenarios
  connectionPoolSize: 64
  failedSlaveReconnectionInterval: 3000
  failedSlaveCheckInterval: 3000

Key tweaks here are extending the command timeout, ensuring enough idle connections are always available, and enabling frequent checks for failed slave nodes.

3. ElastiCache Master-Slave Failover Handling #

Even with a 2-node setup, occasional failovers (planned or unplanned) can disrupt connections. Redisson 3.6.1 has limited support for automatic reconnection to a new master node after failover, which can leave stale connections that trigger timeouts.

Fix:

• Upgrade Redisson to leverage its improved master-slave failover detection (newer versions automatically discover the new master and refresh connections without manual intervention).
• Enable ElastiCache's automatic failover feature if you haven't already, to minimize downtime during node failures.

4. LRU Cache Background Operations Overhead #

The LRU-bounded RMapCache runs background tasks to evict expired entries and trim records beyond your maxSize (500k). In older Redisson versions, these tasks might block connections or consume unexpected resources, leading to timeouts for your main get/put operations—even under low load.

Optimizations:

• Adjust cacheCleanupDelay to set a reasonable interval for background cleanup (e.g., 1000ms instead of the default) to avoid bursty resource usage.
• If you don't need per-entry TTL, disable it to reduce background processing overhead.
• Verify that you're not setting an overly aggressive maxIdleTime that forces frequent, unnecessary evictions.

5. AWS Network Layer Checks #

Even low load can trigger timeouts if there's network misconfiguration:

• Ensure your application and ElastiCache nodes are in the same VPC, and that security groups/network ACLs allow inbound traffic on Redis's default port (6379) from your app servers.
• Check CloudWatch metrics for ElastiCache: look for spikes in ReplicationLag, EngineCPUUtilization, or NetworkLatency which could indicate underlying issues with the cache cluster.
• If your app is in a different AZ than the ElastiCache nodes, consider moving them to the same AZ to reduce network latency.

Final Debugging Steps #

1. Enable debug logging for Redisson to get granular details about timeouts (look for logs about connection attempts, command retries, and node state changes).
2. Test with a single ElastiCache node first to rule out master-slave specific issues.
3. Double-check your RMapCache configuration (confirm maxSize, eviction policies, and TTL settings are correctly applied).

内容的提问来源于stack exchange，提问作者akichidis