Which commands we run
INFO — memory, clients, persistence, replication and keyspace sections
CONFIG GET — maxmemory, eviction policy and persistence settings
SLOWLOG GET / SLOWLOG LEN — slow command records
LATENCY LATEST, LATENCY DOCTOR, LATENCY HISTORY — latency events and diagnostics
MEMORY STATS, MEMORY DOCTOR — memory breakdown and fragmentation
CLIENT LIST — connected clients, idle times and output buffer state
CLUSTER INFO / CLUSTER NODES — cluster state, slot distribution and node roles
ROLE, DBSIZE, TIME — replication role and key counts
What it does specifically for Redis
Memory and eviction analysis
Proximity to maxmemory, fragmentation ratio and eviction rate are tracked together, making it visible at what point eviction starts eroding your hit ratio.
Slow command and latency diagnosis
SLOWLOG records and LATENCY events are combined, correlating latency spikes with forks, AOF rewrites or specific slow commands.
Cluster and replication visibility
Slot distribution, node roles, replica lag and cluster state are tracked, surfacing missing slots or detached replicas.
Minimum privileges required
The baseline is read-only and safe to grant in production.
Prerequisites
·Agentless: the connection originates from the ClusterEye API server; no agent is installed on the target.
·Most of these commands are in the @admin category, so a plain +@read user is not sufficient.
·Redis < 6: no ACL granularity — requirepass only.
·KEYS is never run, key data is never read, and nothing is ever written.