[pull] master from mudler:master

core/application/distributed.go

@@ -41,6 +41,7 @@ type DistributedServices struct {
 	FileStager   nodes.FileStager
 	ModelAdapter *nodes.ModelRouterAdapter
 	Unloader     *nodes.RemoteUnloaderAdapter
+	ModelCleanup *nodes.ModelCleanupService
 
 	shutdownOnce sync.Once
 }
@@ -346,8 +347,10 @@ func initDistributed(cfg *config.ApplicationConfig, authDB *gorm.DB, configLoade
 	if configLoader != nil {
 		conflictResolver = configLoader
 	}
+	modelCleanup := nodes.NewModelCleanupService(registry, remoteUnloader)
 	router := nodes.NewSmartRouter(registry, nodes.SmartRouterOptions{
 		Unloader:         remoteUnloader,
+		ModelCleanup:     modelCleanup,
 		FileStager:       fileStager,
 		GalleriesJSON:    routerGalleriesJSON,
 		AuthToken:        routerAuthToken,
@@ -437,6 +440,7 @@ func initDistributed(cfg *config.ApplicationConfig, authDB *gorm.DB, configLoade
 		FileStager:   fileStager,
 		ModelAdapter: modelAdapter,
 		Unloader:     remoteUnloader,
+		ModelCleanup: modelCleanup,
 	}, nil
 }
 

core/application/startup.go

@@ -298,6 +298,7 @@ func New(opts ...config.AppOption) (*Application, error) {
 		if distSvc.Reconciler != nil {
 			go distSvc.Reconciler.Run(options.Context)
 		}
+		go distSvc.ModelCleanup.Run(options.Context)
 		// In distributed mode, MCP CI jobs are executed by agent workers (not the frontend)
 		// because the frontend can't create MCP sessions (e.g., stdio servers using docker).
 		// The dispatcher still subscribes to jobs.new for persistence (result/progress subs)
@@ -370,13 +371,15 @@ func New(opts ...config.AppOption) (*Application, error) {
 			gs := application.galleryService
 			sys := options.SystemState
 			cfgLoaderOpts := options.ToConfigLoaderOptions()
+			modelRevisionLifecycle := modeladmin.NewDistributedModelRevisionLifecycle(distSvc.Registry, distSvc.ModelCleanup)
+			gs.SetModelRevisionLifecycle(modelRevisionLifecycle)
 			gs.OnModelsChanged = func(evt messaging.CacheInvalidateEvent) {
 				// ApplyRemoteChange honors the op: a "delete" prunes the element
 				// (a reload-from-path is additive and cannot drop it), anything
 				// else reloads from disk; a named element's running instance is
 				// shut down so the new config takes effect. The originating
 				// replica reloads inline and never depends on this path.
-				if err := modeladmin.ApplyRemoteChange(application.ModelConfigLoader(), application.modelLoader, sys.Model.ModelsPath, evt, cfgLoaderOpts...); err != nil {
+				if err := modeladmin.ApplyRemoteChange(options.Context, application.ModelConfigLoader(), sys.Model.ModelsPath, evt, modelRevisionLifecycle, cfgLoaderOpts...); err != nil {
 					xlog.Warn("Failed to apply peer model config change", "error", err)
 				}
 			}

core/backend/ctx_propagation_test.go

@@ -24,8 +24,8 @@ import (
 	"github.com/mudler/LocalAI/core/backend"
 	"github.com/mudler/LocalAI/core/config"
 	"github.com/mudler/LocalAI/core/schema"
-	pbproto "github.com/mudler/LocalAI/pkg/grpc/proto"
 	"github.com/mudler/LocalAI/pkg/distributedhdr"
+	pbproto "github.com/mudler/LocalAI/pkg/grpc/proto"
 	"github.com/mudler/LocalAI/pkg/model"
 	"github.com/mudler/LocalAI/pkg/system"
 
@@ -41,7 +41,7 @@ import (
 func newCapturingLoader() (*model.ModelLoader, *atomic.Value, func() context.Context) {
 	loader := model.NewModelLoader(&system.SystemState{})
 	var captured atomic.Value
-	loader.SetModelRouter(func(ctx context.Context, _ string, _, _, _ string, _ *pbproto.ModelOptions, _ bool) (*model.Model, error) {
+	loader.SetModelRouter(func(ctx context.Context, _ string, _, _, _, _ string, _ *pbproto.ModelOptions, _ bool) (*model.Model, error) {
 		captured.Store(ctx)
 		// Return an error so the backend short-circuits before trying to
 		// dial gRPC. We only care about the context-arrival contract.

core/backend/model_identity_modalities_test.go

@@ -163,7 +163,7 @@ func (r *recordingBackend) VoiceEmbed(_ context.Context, in *pb.VoiceEmbedReques
 // backend, so every helper below reaches it through the real Load path.
 func newRecordingLoader(rec *recordingBackend) *model.ModelLoader {
 	loader := model.NewModelLoader(&system.SystemState{})
-	loader.SetModelRouter(func(_ context.Context, id string, _, _, _ string, _ *pb.ModelOptions, _ bool) (*model.Model, error) {
+	loader.SetModelRouter(func(_ context.Context, id string, _, _, _, _ string, _ *pb.ModelOptions, _ bool) (*model.Model, error) {
 		return model.NewModelWithClient(id, "test://recording", rec), nil
 	})
 	return loader

core/backend/options.go

@@ -202,6 +202,11 @@ func ModelOptions(c config.ModelConfig, so *config.ApplicationConfig, opts ...mo
 		model.WithContext(so.Context),
 		model.WithModelID(c.ModelID()),
 	}
+	if revision, err := config.ModelConfigRevision(&c); err == nil {
+		defOpts = append(defOpts, model.WithConfigRevision(revision))
+	} else {
+		xlog.Warn("Failed to compute model configuration revision", "model", c.ModelID(), "error", err)
+	}
 	managedPrimary := len(c.Artifacts) > 0 && c.Artifacts[0].Resolved != nil
 	if managedPrimary {
 		defOpts = append(defOpts, model.WithModelFile(c.ModelFileName()))

core/config/application_config.go

@@ -1140,6 +1140,7 @@ func (o *ApplicationConfig) ToConfigLoaderOptions() []ConfigLoaderOption {
 		LoadOptionF16(o.F16),
 		LoadOptionThreads(o.Threads),
 		ModelPath(o.SystemState.Model.ModelsPath),
+		LoadOptionGalleryFiles(o.Galleries...),
 	}
 }