redis-sentinel

High availability monitoring for Redis clusters

TLDR

Start Sentinel with a config file

$ redis-sentinel [/etc/redis/sentinel.conf]

Start via redis-server in sentinel mode

$ redis-server [sentinel.conf] --sentinel

List all monitored masters

$ redis-cli -p [26379] sentinel masters

Get current master address by name

$ redis-cli -p [26379] sentinel get-master-addr-by-name [mymaster]

Trigger manual failover

$ redis-cli -p [26379] sentinel failover [mymaster]

SYNOPSIS

redis-sentinel configfile_redis-server configfile_ --sentinel

Redis Sentinel provides high availability for Redis deployments through continuous monitoring, automatic failover, and notification. It watches master and replica instances, and when a master becomes unreachable and a quorum of Sentinel processes agree it is down, it automatically promotes a replica to master and reconfigures the remaining replicas to use the new master.Sentinel also acts as a configuration provider, allowing clients to discover the current master address for a named service. Multiple Sentinel instances (at least three are recommended) form a distributed system that reaches consensus on failover decisions, preventing split-brain scenarios.A configuration file is mandatory; Sentinel rewrites it on failover to persist the new topology.

PARAMETERS

--sentinel

When passed to `redis-server`, starts it in Sentinel mode using the provided config file.

CAVEATS

At least three Sentinel instances are recommended for robust quorum. The default Sentinel port is 26379. Sentinel modifies its configuration file on failover. Sentinel does not shard data — it only monitors and redirects clients.

HISTORY

Redis Sentinel was introduced in Redis 2.4 (stable in 2.8) by Salvatore Sanfilippo to provide high availability without manual intervention.