Every feature of the Wiki works the same on one server or many. What changes is how the servers coordinate.
| Single server | Cluster | |
|---|---|---|
| Licence | Standard or higher | Pro or higher |
| PostgreSQL | Required | Required, shared by all servers |
| Redis | Optional | Required, shared by all servers |
| Storage | Any backend | S3, MinIO, Azure, or a shared volume |
| Servers | 1 | Up to the licence limit |
1. What a cluster shares
- Pages and edits are saved in PostgreSQL first, then announced to the other servers, so two people editing the same page on different servers see each other's changes.
- Background work (imports, analytics rollups, scheduled publishing, review-by reminders, Q&A reputation) runs once per cluster, whichever server picks it up.
- Settings, such as anonymous access, search engines and video sites, apply on every server within seconds of being saved.
- Files go to the shared storage, so any server serves any image or video.
2. Redis
Give Redis as a URL in the setup wizard, the unattended setup file (cache.redis_url) or the recovery page:
| Kind | URL |
|---|---|
| One server | redis://:password@redis.internal:6379/0, or rediss:// for TLS |
| Sentinel group | redis+sentinel://:password@s1:26379?master_name=mymaster&addr=s2:26379&addr=s3:26379 |
| Redis Cluster | redis+cluster://:password@n1:6379?addr=n2:6379&addr=n3:6379 |
- Each kind takes TLS (rediss, rediss+sentinel, rediss+cluster), a private certificate authority (ca_file=/path/ca.pem) and Redis ACL users (user:password@).
- Redis coordinates; PostgreSQL keeps the records. If Redis becomes unavailable, the Wiki keeps working: sign-in and sessions (kept in PostgreSQL), scheduled jobs once per cluster, and live updates through PostgreSQL. Rate limits are then counted by each server on its own. It logs one warning when Redis goes and one message when it returns.
- A failover can lose the last few writes to Redis; results stay correct because PostgreSQL holds the records.
3. Database connections
Each server opens up to pool_max_conns connections to PostgreSQL (by default 4, or the number of CPUs if that is more). Set it in DATABASE_URL, for example postgres://wiki:…@db:5432/wiki?pool_max_conns=25, at a server's first start. In a cluster every server opens its own, so keep servers × pool_max_conns, plus room for maintenance, under PostgreSQL's max_connections.
4. Read cache
Each server keeps what most requests look up, such as roles and settings, in its own memory. PostgreSQL tells every server the moment one of these changes, and each server also checks every second, so a change applies everywhere almost at once, with or without Redis. READ_CACHE_ENABLED=false turns it off on one server (see Environment variables).
5. Clustering page
Settings → Clustering (super admins) lists every server: its status, health, name and machine ID, version, CPU, memory, uptime and last heartbeat, and the leader. A server that stops responding gets a notice above the table.
| Status | Meaning |
|---|---|
| active | Running. |
| draining | Shutting down: finishing its work before it stops. |
| left | Shut down cleanly. Removed from the list after a day (CLUSTER_DEAD_NODE_TTL). |
| dead | Stopped sending heartbeats without shutting down cleanly. It becomes active again by itself with its next heartbeat; one that stays dead is removed after a day. |