feat: allow to run parallel requests (#1290)

* feat: allow to run parallel requests

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* fixup

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

api/backend/options.go

@@ -16,6 +16,10 @@ func modelOpts(c config.Config, o *options.Option, opts []model.Option) []model.
 		opts = append(opts, model.WithSingleActiveBackend())
 	}
 
+	if o.ParallelBackendRequests {
+		opts = append(opts, model.EnableParallelRequests)
+	}
+
 	if c.GRPC.Attempts != 0 {
 		opts = append(opts, model.WithGRPCAttempts(c.GRPC.Attempts))
 	}

api/localai/backend_monitor.go

@@ -125,11 +125,11 @@ func BackendMonitorEndpoint(bm BackendMonitor) func(c *fiber.Ctx) error {
 
 		client := bm.options.Loader.CheckIsLoaded(backendId)
 
-		if client == nil {
+		if client == "" {
 			return fmt.Errorf("backend %s is not currently loaded", backendId)
 		}
 
-		status, rpcErr := client.Status(context.TODO())
+		status, rpcErr := client.GRPC().Status(context.TODO())
 		if rpcErr != nil {
 			log.Warn().Msgf("backend %s experienced an error retrieving status info: %s", backendId, rpcErr.Error())
 			val, slbErr := bm.SampleLocalBackendProcess(backendId)

api/options/options.go

@@ -5,9 +5,9 @@ import (
 	"embed"
 	"encoding/json"
 
+	"github.com/go-skynet/LocalAI/metrics"
 	"github.com/go-skynet/LocalAI/pkg/gallery"
 	model "github.com/go-skynet/LocalAI/pkg/model"
-	"github.com/go-skynet/LocalAI/metrics"
 	"github.com/rs/zerolog/log"
 )
 
@@ -36,7 +36,8 @@ type Option struct {
 
 	AutoloadGalleries bool
 
-	SingleBackend bool
+	SingleBackend           bool
+	ParallelBackendRequests bool
 }
 
 type AppOption func(*Option)
@@ -66,6 +67,10 @@ var EnableSingleBackend = func(o *Option) {
 	o.SingleBackend = true
 }
 
+var EnableParallelBackendRequests = func(o *Option) {
+	o.ParallelBackendRequests = true
+}
+
 var EnableGalleriesAutoload = func(o *Option) {
 	o.AutoloadGalleries = true
 }