Fix: Redis Replication Lag Causing Stale Reads

!Symptoms

• Application reading stale data from Redis replicas
• redis-cli INFO replication shows high master_repl_offset gap
• Replica shows master_link_status:down intermittently
• Cache inconsistencies across application instances

?Root Causes

• Network latency between master and replica (cross-zone or cross-region)
• Replica running out of output buffer due to slow consumption
• Master writing faster than replica can sync (write-heavy workload)
• Replica performing RDB save (BGSAVE) blocking replication
• Large keys causing replication stream to stall

#Diagnosis Steps

1. 1Check replication info: `redis-cli INFO replication` — compare master_repl_offset
2. 2Check replica lag in seconds: `redis-cli INFO replication | grep lag`
3. 3Monitor output buffer: `redis-cli INFO clients | grep output_buffer`
4. 4Check slow log: `redis-cli SLOWLOG GET 10`
5. 5Check memory usage: `redis-cli INFO memory`

>Fix

1. 1Move replica to same availability zone as master to reduce network latency
2. 2Increase client-output-buffer-limit for replica: `config set client-output-buffer-limit 'replica 512mb 128mb 60'`
3. 3Disable RDB persistence on replicas: `config set save ''`
4. 4Break up large keys into smaller keys
5. 5Use WAIT command for critical reads: `WAIT 1 1000` to ensure replication

*Prevention

• Monitor replication lag with Prometheus redis_exporter
• Alert on master_repl_offset divergence > threshold
• Use Redis Sentinel or Cluster for automatic failover
• Read from master for critical consistency-required operations
• Size replicas with same or better specs than master

Related Error Messages