Redis Caching Issues with Multiple Next.js Docker Instances

[HellYeahOmg]

Environment

• Setup: 2 VPS servers running multiple Next.js app instances via Docker-compose
• Redis versions tested: redis-stack, redis-strings
• Framework: Next.js 13 and 14

Problem Summary

Experiencing critical issues with Redis caching that make it unusable in production. Tried both redis-stack and redis-strings with different failure modes.

Issue 1: Redis Stack - Memory Eviction Not Working

Symptoms

• Redis reaches maxmemory limit
• Despite proper eviction policy configuration, keys are not being evicted
• Results in OOM (Out of Memory) errors
• Cache stops working completely

After trying fix redis i decided to try redis-strings, because i could not find out why redis would not evict keys even though they are experied and eviction policy is set. LLMs kept whining about the rejson format...

Issue 2: Redis Strings - Extreme Resource Usage

Symptoms

• Initially works fine
• Later causes severe CPU and memory spikes
• System becomes unstable, all containers restart frequently
• Disabling cache immediately resolves resource usage

Evidence

Screenshot shows clear correlation between cache being enabled and resource spikes

Configuration Details

Questions

1. Are there known issues with Redis caching in multi-instance Docker setups?
2. What's the recommended eviction policy for Next.js applications?
3. Could there be connection pooling issues causing the resource spikes?
4. Are there specific Redis configurations recommended for Next.js apps?