mirror of
https://github.com/mudler/LocalAI.git
synced 2025-12-27 00:19:25 -05:00
fc5b9ebfcc
* feat(loader): refactor single active backend support to LRU This changeset introduces LRU management of loaded backends. Users can set now a maximum number of models to be loaded concurrently, and, when setting LocalAI in single active backend mode we set LRU to 1 for backward compatibility. Signed-off-by: Ettore Di Giacinto <mudler@localai.io> * chore: add tests Signed-off-by: Ettore Di Giacinto <mudler@localai.io> * Update docs Signed-off-by: Ettore Di Giacinto <mudler@localai.io> * Fixups Signed-off-by: Ettore Di Giacinto <mudler@localai.io> --------- Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
93 lines
2.5 KiB
Go
93 lines
2.5 KiB
Go
package application
|
|
|
|
import (
|
|
"context"
|
|
"sync"
|
|
|
|
"github.com/mudler/LocalAI/core/config"
|
|
"github.com/mudler/LocalAI/core/services"
|
|
"github.com/mudler/LocalAI/core/templates"
|
|
"github.com/mudler/LocalAI/pkg/model"
|
|
)
|
|
|
|
type Application struct {
|
|
backendLoader *config.ModelConfigLoader
|
|
modelLoader *model.ModelLoader
|
|
applicationConfig *config.ApplicationConfig
|
|
startupConfig *config.ApplicationConfig // Stores original config from env vars (before file loading)
|
|
templatesEvaluator *templates.Evaluator
|
|
galleryService *services.GalleryService
|
|
agentJobService *services.AgentJobService
|
|
watchdogMutex sync.Mutex
|
|
watchdogStop chan bool
|
|
p2pMutex sync.Mutex
|
|
p2pCtx context.Context
|
|
p2pCancel context.CancelFunc
|
|
agentJobMutex sync.Mutex
|
|
}
|
|
|
|
func newApplication(appConfig *config.ApplicationConfig) *Application {
|
|
return &Application{
|
|
backendLoader: config.NewModelConfigLoader(appConfig.SystemState.Model.ModelsPath),
|
|
modelLoader: model.NewModelLoader(appConfig.SystemState),
|
|
applicationConfig: appConfig,
|
|
templatesEvaluator: templates.NewEvaluator(appConfig.SystemState.Model.ModelsPath),
|
|
}
|
|
}
|
|
|
|
func (a *Application) ModelConfigLoader() *config.ModelConfigLoader {
|
|
return a.backendLoader
|
|
}
|
|
|
|
func (a *Application) ModelLoader() *model.ModelLoader {
|
|
return a.modelLoader
|
|
}
|
|
|
|
func (a *Application) ApplicationConfig() *config.ApplicationConfig {
|
|
return a.applicationConfig
|
|
}
|
|
|
|
func (a *Application) TemplatesEvaluator() *templates.Evaluator {
|
|
return a.templatesEvaluator
|
|
}
|
|
|
|
func (a *Application) GalleryService() *services.GalleryService {
|
|
return a.galleryService
|
|
}
|
|
|
|
func (a *Application) AgentJobService() *services.AgentJobService {
|
|
return a.agentJobService
|
|
}
|
|
|
|
// StartupConfig returns the original startup configuration (from env vars, before file loading)
|
|
func (a *Application) StartupConfig() *config.ApplicationConfig {
|
|
return a.startupConfig
|
|
}
|
|
|
|
func (a *Application) start() error {
|
|
galleryService := services.NewGalleryService(a.ApplicationConfig(), a.ModelLoader())
|
|
err := galleryService.Start(a.ApplicationConfig().Context, a.ModelConfigLoader(), a.ApplicationConfig().SystemState)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
a.galleryService = galleryService
|
|
|
|
// Initialize agent job service
|
|
agentJobService := services.NewAgentJobService(
|
|
a.ApplicationConfig(),
|
|
a.ModelLoader(),
|
|
a.ModelConfigLoader(),
|
|
a.TemplatesEvaluator(),
|
|
)
|
|
|
|
err = agentJobService.Start(a.ApplicationConfig().Context)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
a.agentJobService = agentJobService
|
|
|
|
return nil
|
|
}
|