perf(queue): 按策略动态扩缩异步 Worker
将 River 执行容量按平台模型与用户组的有效并发策略动态调整,并为长任务续租、并发租约原子抢占和限流退避补充保护。\n\n统一平台模型限流继承语义,兼容历史 platformLimits/modelLimits,并为三个迁移图像模型建立独立并发租约。补充管理端显式继承/覆盖配置、指标、单元测试及隔离 PostgreSQL 验收。\n\n验证:go test ./... -count=1;pnpm lint;pnpm test;pnpm build;隔离 PostgreSQL 并发原子性/续租测试;128 任务与三分钟长任务动态 Worker 验收。
This commit is contained in:
@@ -48,6 +48,8 @@ type Config struct {
|
||||
GlobalHTTPProxySource string
|
||||
LogLevel slog.Level
|
||||
BillingEngineMode string
|
||||
AsyncWorkerHardLimit int
|
||||
AsyncWorkerRefreshIntervalSeconds int
|
||||
}
|
||||
|
||||
func Load() Config {
|
||||
@@ -85,13 +87,15 @@ func Load() Config {
|
||||
TaskProgressCallbackURL: env("TASK_PROGRESS_CALLBACK_URL",
|
||||
strings.TrimRight(env("SERVER_MAIN_BASE_URL", "http://localhost:3000"), "/")+"/internal/platform/task-progress-callbacks",
|
||||
),
|
||||
TaskProgressCallbackTimeoutMS: env("TASK_PROGRESS_CALLBACK_TIMEOUT_MS", "5000"),
|
||||
TaskProgressCallbackMaxAttempts: env("TASK_PROGRESS_CALLBACK_MAX_ATTEMPTS", "10"),
|
||||
CORSAllowedOrigin: env("CORS_ALLOWED_ORIGIN", "http://localhost:5178,http://127.0.0.1:5178"),
|
||||
GlobalHTTPProxy: globalProxy.HTTPProxy,
|
||||
GlobalHTTPProxySource: globalProxy.Source,
|
||||
LogLevel: logLevel(env("LOG_LEVEL", "info")),
|
||||
BillingEngineMode: strings.ToLower(env("BILLING_ENGINE_MODE", "observe")),
|
||||
TaskProgressCallbackTimeoutMS: env("TASK_PROGRESS_CALLBACK_TIMEOUT_MS", "5000"),
|
||||
TaskProgressCallbackMaxAttempts: env("TASK_PROGRESS_CALLBACK_MAX_ATTEMPTS", "10"),
|
||||
CORSAllowedOrigin: env("CORS_ALLOWED_ORIGIN", "http://localhost:5178,http://127.0.0.1:5178"),
|
||||
GlobalHTTPProxy: globalProxy.HTTPProxy,
|
||||
GlobalHTTPProxySource: globalProxy.Source,
|
||||
LogLevel: logLevel(env("LOG_LEVEL", "info")),
|
||||
BillingEngineMode: strings.ToLower(env("BILLING_ENGINE_MODE", "observe")),
|
||||
AsyncWorkerHardLimit: envIntValidated("AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT", 2048),
|
||||
AsyncWorkerRefreshIntervalSeconds: envIntValidated("AI_GATEWAY_ASYNC_WORKER_REFRESH_INTERVAL_SECONDS", 5),
|
||||
}
|
||||
}
|
||||
|
||||
@@ -101,6 +105,12 @@ func (c Config) Validate() error {
|
||||
default:
|
||||
return errors.New("BILLING_ENGINE_MODE must be observe, enforce, or hold")
|
||||
}
|
||||
if c.AsyncWorkerHardLimit < 1 || c.AsyncWorkerHardLimit > 10000 {
|
||||
return errors.New("AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT must be between 1 and 10000")
|
||||
}
|
||||
if c.AsyncWorkerRefreshIntervalSeconds < 1 {
|
||||
return errors.New("AI_GATEWAY_ASYNC_WORKER_REFRESH_INTERVAL_SECONDS must be positive")
|
||||
}
|
||||
switch strings.ToLower(strings.TrimSpace(c.IdentitySecretStore)) {
|
||||
case "":
|
||||
case "file":
|
||||
@@ -210,6 +220,18 @@ func envInt(key string, fallback int) int {
|
||||
return parsed
|
||||
}
|
||||
|
||||
func envIntValidated(key string, fallback int) int {
|
||||
value := envValue(key)
|
||||
if value == "" {
|
||||
return fallback
|
||||
}
|
||||
parsed, err := strconv.Atoi(value)
|
||||
if err != nil {
|
||||
return 0
|
||||
}
|
||||
return parsed
|
||||
}
|
||||
|
||||
func logLevel(value string) slog.Level {
|
||||
switch strings.ToLower(value) {
|
||||
case "debug":
|
||||
|
||||
@@ -26,7 +26,7 @@ func TestLoadIdentitySecretStoreUsesNewEnvironmentNamesAndIgnoresLegacyBusinessV
|
||||
}
|
||||
|
||||
func TestValidateIdentityFileSecretStoreRequiresDirectory(t *testing.T) {
|
||||
cfg := Config{IdentitySecretStore: "file"}
|
||||
cfg := Config{IdentitySecretStore: "file", AsyncWorkerHardLimit: 2048, AsyncWorkerRefreshIntervalSeconds: 5}
|
||||
if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "IDENTITY_SECRET_DIR") {
|
||||
t.Fatalf("Validate() error = %v, want missing identity secret directory", err)
|
||||
}
|
||||
@@ -37,7 +37,12 @@ func TestValidateIdentityFileSecretStoreRequiresDirectory(t *testing.T) {
|
||||
}
|
||||
|
||||
func TestValidateIdentityKubernetesSecretStore(t *testing.T) {
|
||||
cfg := Config{IdentitySecretStore: "kubernetes", IdentityKubernetesSecretName: "easyai-gateway-identity"}
|
||||
cfg := Config{
|
||||
IdentitySecretStore: "kubernetes",
|
||||
IdentityKubernetesSecretName: "easyai-gateway-identity",
|
||||
AsyncWorkerHardLimit: 2048,
|
||||
AsyncWorkerRefreshIntervalSeconds: 5,
|
||||
}
|
||||
if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "namespace") {
|
||||
t.Fatalf("Validate() error = %v, want missing namespace", err)
|
||||
}
|
||||
@@ -52,8 +57,27 @@ func TestValidateIdentitySecurityEventTiming(t *testing.T) {
|
||||
IdentitySecurityEventHeartbeatIntervalSeconds: 60,
|
||||
IdentitySecurityEventStaleAfterSeconds: 60,
|
||||
IdentitySecurityEventClockSkewSeconds: 60,
|
||||
AsyncWorkerHardLimit: 2048,
|
||||
AsyncWorkerRefreshIntervalSeconds: 5,
|
||||
}
|
||||
if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "heartbeat") {
|
||||
t.Fatalf("Validate() error = %v, want invalid stale threshold", err)
|
||||
}
|
||||
}
|
||||
|
||||
func TestValidateAsyncWorkerSettings(t *testing.T) {
|
||||
cfg := Config{AsyncWorkerHardLimit: 10001, AsyncWorkerRefreshIntervalSeconds: 5}
|
||||
if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "HARD_LIMIT") {
|
||||
t.Fatalf("Validate() error = %v, want invalid hard limit", err)
|
||||
}
|
||||
cfg.AsyncWorkerHardLimit = 2048
|
||||
cfg.AsyncWorkerRefreshIntervalSeconds = 0
|
||||
if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "REFRESH_INTERVAL") {
|
||||
t.Fatalf("Validate() error = %v, want invalid refresh interval", err)
|
||||
}
|
||||
t.Setenv("AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT", "not-an-integer")
|
||||
loaded := Load()
|
||||
if err := loaded.Validate(); err == nil || !strings.Contains(err.Error(), "HARD_LIMIT") {
|
||||
t.Fatalf("Validate() error = %v, want invalid non-integer hard limit", err)
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user