use SINGLE_ACTIVE_BACKEND for testing AIO?

Signed-off-by: Dave Lee <dave@gray101.com>

tests/e2e-aio/e2e_suite_test.go

@@ -1,17 +1,12 @@
 package e2e_test
 
 import (
-	"bytes"
 	"context"
-	"encoding/json"
 	"fmt"
-	"io"
-	"net/http"
 	"os"
 	"runtime"
 	"testing"
 
-	"github.com/mudler/LocalAI/core/schema"
 	. "github.com/onsi/ginkgo/v2"
 	. "github.com/onsi/gomega"
 	"github.com/ory/dockertest/v3"
@@ -76,40 +71,6 @@ var _ = AfterEach(func() {
 	//Expect(dbClient.Clear()).To(Succeed())
 })
 
-func ShutdownModel(modelName string) func() {
-	return func() {
-		req := schema.BackendMonitorRequest{
-			BasicModelRequest: schema.BasicModelRequest{
-				Model: modelName,
-			},
-		}
-		serialized, err := json.Marshal(req)
-		Expect(err).To(BeNil())
-		Expect(serialized).ToNot(BeNil())
-
-		// r1, err := http.Post(apiEndpoint+"/backend/monitor", "application/json", bytes.NewReader(serialized))
-		// Expect(err).To(BeNil())
-		// Expect(r1).ToNot(BeNil())
-		// b1, err := io.ReadAll(r1.Body)
-		// GinkgoWriter.Printf("TEMPORARY MONITOR RESPONSE: %q\n", b1)
-
-		resp, err := http.Post(apiEndpoint+"/backend/shutdown", "application/json", bytes.NewReader(serialized))
-		Expect(err).To(BeNil())
-		Expect(resp).ToNot(BeNil())
-		body, err := io.ReadAll(resp.Body)
-		Expect(err).ToNot(HaveOccurred())
-		// if a test fails to load the model, we will recieve a 500 error when we try to shut it down.
-		// We therefore handle both cases seperately
-		if resp.StatusCode == 500 {
-			body, err := io.ReadAll(resp.Body)
-			Expect(err).To(BeNil())
-			Expect(body).To(ContainSubstring(fmt.Sprintf("%s not found", modelName)), fmt.Sprintf("unexpected response during shutdown: %q", body))
-		} else {
-			Expect(resp.StatusCode).To(Equal(200), fmt.Sprintf("failed to shutdown model: %s, response: %+v", body, resp))
-		}
-	}
-}
-
 func startDockerImage() {
 	p, err := dockertest.NewPool("")
 	Expect(err).To(Not(HaveOccurred()))
@@ -134,7 +95,7 @@ func startDockerImage() {
 		PortBindings: map[docker.Port][]docker.PortBinding{
 			"8080/tcp": []docker.PortBinding{{HostPort: apiPort}},
 		},
-		Env:    []string{"MODELS_PATH=/models", "DEBUG=true", "THREADS=" + fmt.Sprint(proc)},
+		Env:    []string{"MODELS_PATH=/models", "DEBUG=true", "THREADS=" + fmt.Sprint(proc), "LOCALAI_SINGLE_ACTIVE_BACKEND=true"},
 		Mounts: []string{md + ":/models"},
 	}
 

tests/e2e-aio/e2e_test.go

@@ -163,7 +163,6 @@ var _ = Describe("E2E test", func() {
 						Model: openai.AdaEmbeddingV2,
 					},
 				)
-				DeferCleanup(ShutdownModel(string(openai.AdaEmbeddingV2)))
 				Expect(err).ToNot(HaveOccurred())
 				Expect(len(resp.Data)).To(Equal(1), fmt.Sprint(resp))
 				Expect(resp.Data[0].Embedding).ToNot(BeEmpty())
@@ -193,7 +192,6 @@ var _ = Describe("E2E test", func() {
 								},
 							},
 						}})
-				DeferCleanup(ShutdownModel(model))
 				Expect(err).ToNot(HaveOccurred())
 				Expect(len(resp.Choices)).To(Equal(1), fmt.Sprint(resp))
 				Expect(resp.Choices[0].Message.Content).To(Or(ContainSubstring("wooden"), ContainSubstring("grass")), fmt.Sprint(resp.Choices[0].Message.Content))
@@ -206,7 +204,6 @@ var _ = Describe("E2E test", func() {
 					Input: "Hello!",
 					Voice: openai.VoiceAlloy,
 				})
-				DeferCleanup(ShutdownModel(string(openai.TTSModel1)))
 				Expect(err).ToNot(HaveOccurred())
 				defer res.Close()
 
@@ -227,15 +224,13 @@ var _ = Describe("E2E test", func() {
 					FilePath: file,
 				}
 				resp, err := client.CreateTranscription(context.Background(), req)
-				DeferCleanup(ShutdownModel(string(openai.Whisper1)))
 				Expect(err).ToNot(HaveOccurred())
 				Expect(resp.Text).To(ContainSubstring("This is the"), fmt.Sprint(resp.Text))
 			})
 		})
 		Context("vad", func() {
 			It("correctly", func() {
 				modelName := "silero-vad"
-				DeferCleanup(ShutdownModel(modelName))
 				req := schema.VADRequest{
 					BasicModelRequest: schema.BasicModelRequest{
 						Model: modelName,
@@ -268,7 +263,6 @@ var _ = Describe("E2E test", func() {
 		Context("reranker", func() {
 			It("correctly", func() {
 				modelName := "jina-reranker-v1-base-en"
-				DeferCleanup(ShutdownModel(modelName))
 
 				req := schema.JINARerankRequest{
 					BasicModelRequest: schema.BasicModelRequest{