TheDocumentation Index
Fetch the complete documentation index at: https://mintlify.com/wikioasis/salt/llms.txt
Use this file to discover all available pages before exploring further.
redis Salt state installs redis-server, renders a Jinja-templated redis.conf from pillar data, and keeps the service running and restarted whenever the configuration changes. Each redis* VM in the fleet carries a host-specific pillar file that overrides the maxmemory limit and eviction policy, while all other tunables inherit cluster-wide defaults from the base redis pillar.
State overview
Configuration file
Renders
/etc/redis/redis.conf from salt://redis/files/redis.conf.jinja, owned by root:redis, mode 0640. All values are sourced from the redis pillar key.Configuration template
The template is intentionally minimal — every directive maps directly to a pillar key with a safe default, making it easy to override individual settings per host.Pillar reference
All keys sit under the top-levelredis mapping. The cluster-wide defaults are defined in pillar/redis/init.sls and selectively overridden in per-host pillar files.
| Key | Default | Description |
|---|---|---|
bind | 127.0.0.1 | Interface(s) Redis listens on; fleet uses 0.0.0.0 for internal reachability |
port | 6379 | TCP port |
protected_mode | yes | Reject connections from non-loopback clients if no password set; fleet sets no |
timeout | 0 | Idle connection timeout in seconds; 0 disables |
tcp_keepalive | 300 | TCP keepalive interval in seconds |
loglevel | notice | Logging verbosity (debug, verbose, notice, warning) |
logfile | "" | Log file path; empty string sends logs to stdout / journald |
databases | 16 | Number of logical databases |
maxmemory | 0 | Memory cap; 0 = unlimited. Use gb/mb suffix (e.g. 12gb) |
maxmemory_policy | noeviction | Eviction policy when maxmemory is reached |
requirepass | (unset) | Optional password; omit the key entirely to disable AUTH |
appendonly | no | Enable AOF persistence (yes/no) |
appendfsync | everysec | AOF fsync strategy (always, everysec, no) |
save | ['900 1', '300 10', '60 10000'] | RDB save checkpoints as a list of "seconds changes" strings |
The
maxmemory_policy for all MediaWiki caching instances is set to allkeys-lru, which evicts the least-recently-used key from the entire keyspace when the memory limit is reached. This is the correct policy for object caches where every key is expendable.Pillar examples
- Cluster defaults (init.sls)
- redis-us-east-011 (12 GB)
- redis-us-east-012 (3 GB)
- redis-us-east-021 (3 GB)
Fleet topology
redis-us-east-011
Primary US-East cache node. 12 GB memory cap,
allkeys-lru eviction. VM ID 140 on metal-us-east-01.redis-us-east-012
Secondary US-East cache node. 3 GB memory cap,
allkeys-lru eviction. VM ID 141 on metal-us-east-01.redis-us-east-021
US-East-02 cache node. 3 GB memory cap,
allkeys-lru eviction. Hosted on metal-us-east-02 at 10.0.2.90.