Files
LocalAI/core/services/nodes/model_router.go
LocalAI [bot] 06e777b75e feat(distributed): gated X-LocalAI-Node response header (middleware + wrapper) (#9976)
* feat(distributed): add per-request node ID context holder

Introduce pkg/distributedhdr, a leaf package carrying a per-request
*atomic.Value holder for the picked worker node ID from the
SmartRouter (core/services/nodes) up to the HTTP response writer
wrapper (core/http/middleware). Avoids the import cycle that a shared
key in either consumer would create.

Exposes NewHolder, WithHolder, Holder, Stamp, Load, Inherit. The
holder is atomic.Value so cross-goroutine publish from the router to
the response writer wrapper is race-clean.

Assisted-by: Claude:claude-opus-4-7[1m]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* feat(distributed): add ExposeNodeHeader middleware + response writer wrapper

New ApplicationConfig.ExposeNodeHeader bool + --expose-node-header CLI
flag / LOCALAI_EXPOSE_NODE_HEADER env var (default off; the node ID
reveals internal topology and is opt-in).

The middleware creates a per-request *atomic.Value holder, attaches it
to c.Request().Context() via distributedhdr.WithHolder, and wraps
c.Response().Writer with a custom http.ResponseWriter that sets the
X-LocalAI-Node header on first Write / WriteHeader / Flush by reading
the holder. Implements http.Flusher, http.Hijacker, Unwrap so it
composes cleanly with Echo and http.NewResponseController.

request.go propagates the holder onto derived contexts via
distributedhdr.Inherit so the holder survives the correlation-ID
context replacement.

Unit + race-clean concurrency + integration specs.

Assisted-by: Claude:claude-opus-4-7[1m]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* feat(distributed): stamp node ID in router and wire middleware to inference routes

ModelRouterAdapter.Route stamps the picked node ID into the
per-request holder via distributedhdr.Stamp(ctx, result.Node.ID) right
after replica selection.

Wire ExposeNodeHeader middleware to:
- OpenAI chat/completion/embeddings + audio transcriptions/speech + image generations/inpainting
- Anthropic /v1/messages
- Ollama /api/chat, /api/generate, /api/embed, /api/embeddings
- Jina /v1/rerank
- LocalAI /v1/vad

The middleware's wrapper reads the holder on first byte and sets the
X-LocalAI-Node response header before delegating to the underlying
writer. Per-request scope means no race under concurrent multi-replica
routing.

Assisted-by: Claude:claude-opus-4-7[1m]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* fix(distributed): thread request context through backend Load + cover ctx propagation

Five non-OpenAI backend helpers were silently using app.Context instead
of the request context for the gRPC backend call: transcription, TTS,
image generation, rerank, VAD. Effect: distributedhdr.Stamp in the
router callback was a silent no-op for these paths, AND client
cancellation didn't propagate to in-flight inference.

Thread c.Request().Context() (or the equivalent input.Context after
the request middleware has installed the correlation-ID derived
context) through each helper and into ModelOptions via
model.WithContext(ctx). ImageGeneration's signature gains a leading
ctx parameter; in-tree callers (openai image, openai inpainting,
openai inpainting_test) are updated to match.

ModelEmbedding gains a leading ctx parameter for the same reason; the
openai and ollama embedding handlers pass the request context through.

chat_stream_workers.go defers the initial role=assistant chunk
emission until the first token callback so the wrapper's lazy
X-LocalAI-Node lookup against the loader runs AFTER ml.Load has
stamped the per-modelID node ID; semantically identical for clients
(role still arrives before any text).

Regression test core/backend/ctx_propagation_test.go pins ctx
propagation for all five helpers.

Docs updated to enumerate the full endpoint coverage of the
--expose-node-header flag.

Assisted-by: Claude:claude-opus-4-7[1m]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Co-authored-by: Ettore Di Giacinto <mudler@localai.io>
2026-05-25 10:51:48 +02:00

106 lines
3.7 KiB
Go

package nodes
import (
"context"
"fmt"
"sync"
"github.com/mudler/LocalAI/pkg/distributedhdr"
pb "github.com/mudler/LocalAI/pkg/grpc/proto"
"github.com/mudler/LocalAI/pkg/model"
"github.com/mudler/xlog"
)
// ModelRouterAdapter wraps SmartRouter to provide a model.ModelRouter callback
// for the ModelLoader. When the ModelLoader needs to start a gRPC backend,
// it calls this adapter instead of starting a local process.
//
// The adapter:
// 1. Calls SmartRouter.Route() to find/load the model on a remote node
// (SmartRouter pre-stages model files via FileStager in Route())
// 2. Returns a Model with a FileStagingClient-wrapped gRPC client
// 3. Tracks Release() functions for cleanup on model unload
type ModelRouterAdapter struct {
router *SmartRouter
mu sync.Mutex
release map[string]func() // modelID -> Release() callback
}
// NewModelRouterAdapter creates a new adapter.
func NewModelRouterAdapter(router *SmartRouter) *ModelRouterAdapter {
return &ModelRouterAdapter{
router: router,
release: make(map[string]func()),
}
}
// Route implements the model.ModelRouter callback signature.
// It delegates to SmartRouter.Route() and returns a Model that wraps the
// remote gRPC client with file staging if configured.
func (a *ModelRouterAdapter) Route(ctx context.Context, backend, modelID, modelName, modelFile string,
opts *pb.ModelOptions, parallel bool) (*model.Model, error) {
backendType := backend
// Set model file and name on opts — these are passed as function args
// (like the local path in initializers.go) but not pre-set in gRPCOptions.
if opts != nil {
opts.Model = modelName
opts.ModelFile = modelFile
}
// Route to a remote node (SmartRouter handles model pre-staging via FileStager)
// Pass modelID so the DB tracks models by their logical ID, not the file path
result, err := a.router.Route(ctx, modelID, modelName, backendType, opts, parallel)
if err != nil {
return nil, fmt.Errorf("routing model %s: %w", modelName, err)
}
// Store release function for cleanup on unload
a.mu.Lock()
if oldRelease, ok := a.release[modelID]; ok {
go oldRelease() // clean up previous route in background
}
a.release[modelID] = result.Release
a.mu.Unlock()
// The client from Route is already a grpc.Backend.
// If file staging is configured, it's already wrapped with FileStagingClient
// by SmartRouter. Use NewModelWithClient so the wrapper is preserved when
// the ModelLoader returns this model on subsequent requests.
m := model.NewModelWithClient(modelID, result.Node.Address, result.Client)
// Publish the picked node ID into the per-request holder attached to
// ctx (by middleware.ExposeNodeHeader). No-op when the holder is
// absent - e.g. health-check probes, watchdog reloads or any path that
// did not flow through the HTTP middleware. The response writer
// wrapper reads this on the first byte to set X-LocalAI-Node, so
// attribution is exact for the request that triggered THIS Route call,
// even when several requests for the same modelID are being routed
// concurrently to different replicas.
distributedhdr.Stamp(ctx, result.Node.ID)
xlog.Info("Model routed to remote node", "model", modelName, "node", result.Node.Name, "address", result.Node.Address)
return m, nil
}
// ReleaseModel releases the in-flight counter for a model.
// Called when the model is unloaded.
func (a *ModelRouterAdapter) ReleaseModel(modelID string) {
a.mu.Lock()
release, ok := a.release[modelID]
if ok {
delete(a.release, modelID)
}
a.mu.Unlock()
if ok && release != nil {
release()
}
}
// AsModelRouter returns a model.ModelRouter function suitable for ModelLoader.SetModelRouter().
func (a *ModelRouterAdapter) AsModelRouter() model.ModelRouter {
return a.Route
}