Redis (cache / queue)

Session store for jarvisos-auth and task queue for the control plane.

Class

tier-0 Fleet-fatal: loss stops the estate or risks data. Everything else waits.

RTO (declared)

15m

RTO (achievable)

30m set by home1

RPO

—

State

rebuildable

Redundancy

single node

Failover

none

Failure domain

fd-home-rack

Health check: TCP probe every 60s — proves the listener accepts connections. Observed independently of the service.
Blind spot: Does not check that the queue is draining. A wedged consumer with a healthy Redis looks fine here.

Without it: Every session is invalidated: nobody is logged in and no published route authorises until Redis is back. Queued work re-runs.

• § Accepted risk: The class requires redundancy and there is none. redis
  Redis (cache / queue) is single-node but tier-0 requires a redundancy model.
  Accepted by homelab-operator, review 2026-11-15: Contents are rebuildable and a restart is seconds. A Sentinel cluster needs a third voting domain this estate does not have, and would introduce split-brain risk to buy availability for data we are willing to lose. The exposure is the auth coupling below, not the data.

Full classification and SLOs →