cache
Distributed caching with TTL, atomic operations, pattern matching, and multi-tenant isolation. Provides in-memory (development/tests) and Redis (production) backends.
Key namespaces
| Namespace | Purpose |
|---|---|
|
Protocols: |
|
Malli schemas for |
|
In-memory adapter (atoms, LRU eviction, TTL) |
|
Redis adapter (Jedis, connection pooling, JSON serialization) |
|
Tenant-scoped wrapper with automatic key prefixing |
Usage
(require '[boundary.cache.shell.adapters.in-memory :as in-mem]
'[boundary.cache.ports :as ports])
;; Create in-memory cache
(def cache (in-mem/create-in-memory-cache {:max-size 1000 :track-stats? true}))
;; Basic operations
(ports/set-value! cache :user-123 {:name "Alice"} 3600) ; TTL in seconds
(ports/get-value cache :user-123)
(ports/delete-value! cache :user-123)
;; Atomic operations
(ports/increment! cache :counter)
(ports/set-if-absent! cache :lock "holder-id" 30) ; SETNX pattern
;; Tenant-scoped cache
(require '[boundary.cache.shell.tenant-cache :as tenant-cache])
(def tenant-cache (tenant-cache/create-tenant-cache cache "acme-corp"))
;; Keys automatically prefixed: "tenant:acme-corp:user-123"Conventions
-
Keys can be strings or keywords; internally converted to strings
-
TTL is in seconds;
nilfromttl()means no expiration -
In-memory stores Clojure values as-is; Redis uses JSON serialization
-
flush-all!on a tenant cache only deletes that tenant’s keys (safe)
Important notes
-
Pattern matching is O(n) for in-memory (regex scan); Redis uses cursor-based SCAN
-
Redis adapter uses JSON — Java objects and Clojure functions cannot be stored
-
Do not share an OSM geo-adapter across threads (rate-limiting atom is per-instance)
Integrant configuration
;; In-memory (development)
:boundary/cache {:backend :in-memory :max-size 10000}
;; Redis (production)
:boundary/cache {:backend :redis :url #env REDIS_URL :pool-size 10}Testing
clojure -M:test:db/h2 :cache