Redis Data Source

Redis data source is suitable for low-latency identity lookups and cache-oriented architectures.

Suitable Scenarios

• Identity and ACL data are already modeled in Redis.
• You need lightweight integration with existing cache systems.
• You want fast sync source behavior with simple key/hash contracts.

Core Capabilities

• Read user and ACL data using agreed key/hash patterns.
• Keep auth hot path memory-based inside broker.
• Reuse existing Redis structures without mandatory schema migration.

Runtime Model (Brief)

1. Broker syncs auth-related data from Redis.
2. Synced data updates local cache.
3. CONNECT auth checks in-memory cache first.

Configuration

Key fields in redis_config:

• redis_addr: Redis endpoint (for example 127.0.0.1:6379)
• mode: Single / Cluster / Sentinel
• database: DB index
• password: Redis password
• query_user: command to fetch user ID list (for example SMEMBERS mqtt:users)
• query_acl: command to fetch ACL ID list (for example SMEMBERS mqtt:acls)
• query_blacklist: command to fetch blacklist ID list (for example SMEMBERS mqtt:blacklists)

The adapter first executes query_user/query_acl/query_blacklist to get ID lists, then reads hash details by key convention (for example mqtt:user:{id}, mqtt:acl:{id}, mqtt:blacklist:{id}).

Hash Field Contract

User hash (mqtt:user:{username})

Required:

• password
• is_superuser (1 or 0)

Optional:

• salt
• created (unix seconds)

ACL hash (mqtt:acl:{id})

Required:

• permission (1=Allow, 0=Deny)
• access (0..5)

Optional:

• username
• clientid
• ipaddr
• topic

Blacklist hash (mqtt:blacklist:{id})

Required:

• blacklist_type
• resource_name
• end_time

Optional:

• desc

Example

[[mqtt.auth]]
authn_type = "password_based"

[mqtt.auth.config.storage_config]
storage_type = "redis"

[mqtt.auth.config.storage_config.redis_config]
redis_addr = "127.0.0.1:6379"
mode = "Single"
database = 0
password = ""
query_user = "SMEMBERS mqtt:users"
query_acl = "SMEMBERS mqtt:acls"
query_blacklist = "SMEMBERS mqtt:blacklists"