Redis down

What does this alert mean?

redis_up == 0. Redis is not responding to the exporter.

Contributing factors

• Process crashed or was OOM-killed
• maxmemory reached with policy noeviction
• AOF or RDB write failure causing crash

Affected components

• All services using Redis for caching, sessions, or queues

Expected action

Restore the Redis process. If maxmemory is the cause, review eviction policy before restarting.

Metric

Description & unit

Binary gauge from redis_exporter. 1 = reachable, 0 = down.

Unit: boolean (0 or 1)

Threshold rationale

Any 0 means Redis is unreachable.

Normal behavior

Always 1.

Screenshots

Silencing

Silence during planned maintenance.

Expected frequency

Rare.

Severity guidance

Critical. Impact depends on whether Redis is used for caching only (degraded performance) or for sessions/queues (data loss risk).

Impacted users

All users of services backed by this Redis instance.

Things to check to determine severity

• Is Redis used for sessions? If so, users may be logged out.
• Is Redis used as a queue? Check for message loss.

Prometheus query

Dashboards

• Redis Overview — Fill in: dashboard URL

Log queries

• Redis logs — Fill in: log query URL

Steps

1. 1Check Redis logs: /var/log/redis/redis-server.log
2. 2Check process: systemctl status redis
3. 3Check system memory: free -h
4. 4Check Redis maxmemory config: redis-cli CONFIG GET maxmemory
5. 5Attempt restart: systemctl restart redis

Useful commands

Guidance

Escalate immediately if Redis is used for sessions or job queues.

Slack channels

• #infrastructure-oncall — Fill in: channel URL

Alert rule

Tuning notes

Add `for: 1m` to tolerate a single restart.