Open Source|All improvements go through Pull Requests with automated CI checks before review by@Leo-Galli|Join Discord|Contributing guide
Deployment topologies

Architecture — Deployment

Deployment topologies, failure domains and scaling.

Single host (recommended)

One Docker host runs everything via docker compose. Backups are volume snapshots + pg_dump.

Split tier

For fleets above a few hundred servers:

  • Web + API behind a load balancer (2+ replicas).
  • Workers on dedicated hosts scaled by queue depth.
  • PostgreSQL on managed provider or dedicated VM with PITR.
  • Redis with AOF persistence and a replica.

Multi-region

Each region owns its nodes, PostgreSQL and Redis. Web tier served from CDN. Cross-region features built on webhook and sync queues.

Failure domains

FailureImpactMitigation
Web app downPortal unreachableMultiple replicas, health checks
API downAll operations failReplicas, /health probes
Worker downQueues pile upRestart policy, queue lag alert
PostgreSQL downRead/write failsPITR backups, managed provider
Redis downQueues and cache lostAOF persistence, replica

Data model

See also: Architecture overview, Monitoring.