gas-cache

Cache service for the Gas ecosystem. Provides two gas.CacheProvider implementations — an in-memory backend for development and testing, and a Valkey (Redis-compatible) backend for production.

Install

go get github.com/gasmod/gas-cache

Backends

Backend	Package	Use case
In-memory	github.com/gasmod/gas-cache/memory	Development, testing, single-instance deployments
Valkey (Redis)	github.com/gasmod/gas-cache/valkey	Production, multi-instance deployments

Both backends implement gas.Service and gas.CacheProvider. The Valkey backend also implements gas.HealthReporter and gas.ReadyReporter for Kubernetes-style liveness and readiness probes.

Usage

In-memory backend

package main

import (
	"github.com/gasmod/gas"
	cachemem "github.com/gasmod/gas-cache/memory"
)

func main() {
	app := gas.NewApp(
		gas.WithSingletonService[*cachemem.Service](cachemem.New()),
		// ...
	)

	app.Run()
}

With custom configuration:

cfg := cachemem.DefaultConfig()
cfg.Cache.MaxEntries = 10000
cfg.Cache.DefaultTTL = 5 * time.Minute
cfg.Cache.CleanupInterval = 2 * time.Minute

cachemem.New(cachemem.WithConfig(cfg))

Valkey backend

package main

import (
	"github.com/gasmod/gas"
	cachevk "github.com/gasmod/gas-cache/valkey"
)

func main() {
	app := gas.NewApp(
		gas.WithSingletonService[*cachevk.Service](cachevk.New()),
		// ...
	)

	app.Run()
}

With custom configuration:

cfg := cachevk.DefaultConfig()
cfg.Cache.Addr = "valkey.internal:6379"
cfg.Cache.Password = "secret"
cfg.Cache.DB = 1
cfg.Cache.ConnRetries = 3

cachevk.New(cachevk.WithConfig(cfg))

Dependency injection

Services receive the cache through gas.CacheProvider via constructor injection:

type Service struct {
	cache gas.CacheProvider
}

func New(cache gas.CacheProvider) *Service {
	return &Service{cache: cache}
}

func (s *Service) Init() error {
	ctx := context.Background()
	_ = s.cache.Set(ctx, "hello", []byte("world"), 5*time.Minute)
	return nil
}

Direct Valkey client access

For advanced Valkey operations beyond the CacheProvider interface, type-assert to access the underlying client:

type ValkeyProvider interface {
	Client() valkey.Client
}

func (s *Service) Init() error {
	if vp, ok := s.cache.(ValkeyProvider); ok {
		client := vp.Client()
		// use client for pub/sub, Lua scripts, etc.
	}
	return nil
}

Config

If WithConfig is not provided, both backends automatically bind configuration from the gas.ConfigProvider injected via DI. This lets you drive cache settings from environment variables or a config file without any explicit wiring.

Memory config

Field	Default	Description
Cache.MaxEntries	0	Max entries (0 = unlimited)
Cache.CleanupInterval	1m	How often expired entries are evicted (0 = disabled)
Cache.DefaultTTL	0	Default TTL when Set is called with ttl=0 (0 = never expires)

Valkey config

Field	Default	Description
Cache.Addr	localhost:6379	Valkey server address (host:port)
Cache.Password		Authentication password (empty = no auth)
Cache.DB	0	Database number (0-15)
Cache.DialTimeout	5s	Timeout for establishing new connections
Cache.WriteTimeout	3s	Timeout for write operations and periodic PINGs
Cache.ConnRetries	0	Number of connection retry attempts (0 = no retries)
Cache.ConnRetryInterval	2s	Base retry interval; doubles each attempt (exp. backoff)

Testing

The cachetest package provides a mock implementation of gas.CacheProvider:

import "github.com/gasmod/gas-cache/cachetest"

mock := &cachetest.MockCache{}
mock.GetFn = func(ctx context.Context, key string) ([]byte, error) {
	return []byte("hello"), nil
}

// pass mock as gas.CacheProvider
// assert calls:
if mock.CallCount("Get") != 1 {
	t.Error("expected one Get call")
}

Health and Readiness

The Valkey backend implements gas.HealthReporter and gas.ReadyReporter:

• CheckHealth(ctx) — liveness. Returns cache.ErrClosed once the service is closed and nil otherwise. The valkey-go client reconnects internally, so a transient network failure is not a liveness failure — a process restart would not help.
• CheckReady(ctx) — readiness. Issues a PING against the Valkey server using the caller's context. Returns an error while the dependency is unreachable so traffic can be drained until it recovers.

The in-memory backend intentionally does not implement these interfaces: it has no external dependency and no warmup, so liveness and readiness track the service lifecycle that the gas framework already manages.

Sentinel Errors

The root cache package defines two sentinel errors used by both backends:

cache.ErrKeyNotFound // returned by Get when the key does not exist or has expired
cache.ErrClosed      // returned when an operation is attempted on a closed service