fix(tests): make e2e backend model downloads resumable and stall-based

Makefile

@@ -645,6 +645,9 @@ test-extra: prepare-test-extra
 ## suite against it.
 ##
 BACKEND_TEST_MODEL_URL?=https://huggingface.co/Qwen/Qwen3-0.6B-GGUF/resolve/main/Qwen3-0.6B-Q8_0.gguf
+## Suite timeout for `go test`. Wrappers whose model download alone can eat
+## most of the default (multi-GB models on a slow HF CDN day) override this.
+BACKEND_TEST_TIMEOUT?=30m
 
 ## Generic target — runs the suite against whatever BACKEND_IMAGE points at.
 ## Depends on protogen-go so pkg/grpc/proto is generated before `go test`.
@@ -672,7 +675,7 @@ test-extra-backend: protogen-go
 	BACKEND_TEST_FACE_IMAGE_3_URL="$$BACKEND_TEST_FACE_IMAGE_3_URL" \
 	BACKEND_TEST_FACE_IMAGE_3_FILE="$$BACKEND_TEST_FACE_IMAGE_3_FILE" \
 	BACKEND_TEST_VERIFY_DISTANCE_CEILING="$$BACKEND_TEST_VERIFY_DISTANCE_CEILING" \
-	go test -v -timeout 30m ./tests/e2e-backends/...
+	go test -v -timeout $(BACKEND_TEST_TIMEOUT) ./tests/e2e-backends/...
 
 ## Convenience wrappers: build the image, then exercise it.
 test-extra-backend-llama-cpp: docker-build-llama-cpp
@@ -1012,6 +1015,7 @@ test-extra-backend-vibevoice-cpp-tts: docker-build-vibevoice-cpp
 ## post-image disk budget.
 test-extra-backend-vibevoice-cpp-transcription: docker-build-vibevoice-cpp
 	BACKEND_IMAGE=local-ai-backend:vibevoice-cpp \
+	BACKEND_TEST_TIMEOUT=120m \
 	BACKEND_TEST_MODEL_URL='https://huggingface.co/mudler/vibevoice.cpp-models/resolve/main/vibevoice-asr-q4_k.gguf#vibevoice-asr-q4_k.gguf' \
 	BACKEND_TEST_EXTRA_FILES='https://huggingface.co/mudler/vibevoice.cpp-models/resolve/main/tokenizer.gguf#tokenizer.gguf' \
 	BACKEND_TEST_AUDIO_URL=https://github.com/ggml-org/whisper.cpp/raw/master/samples/jfk.wav \

tests/e2e-backends/backend_test.go

@@ -1453,20 +1453,34 @@ func extractImage(image, dest string) {
 
 // downloadFile fetches url into dest using curl -L. Used for CI convenience;
 // local runs can use BACKEND_TEST_MODEL_FILE to skip downloading.
-// Retry flags guard against transient CI network hiccups (github.com in
-// particular has been flaky from GHA runners, timing out TCP connects).
+// Model files can be huge (the vibevoice ASR e2e model is >10 GB), so a
+// wall-clock cap per attempt (--max-time) is the wrong guard: on a slow CDN
+// day no attempt can ever finish, and curl's --retry restarts from byte
+// zero, so repeated attempts make no forward progress (this hung the
+// vibevoice transcription job until the suite timeout). Instead, abort only
+// on a real stall (--speed-limit/--speed-time) and resume from the bytes
+// already on disk (-C -), with the retry loop out here because curl does
+// not re-evaluate the resume offset on its internal retries.
 func downloadFile(url, dest string) {
 	GinkgoHelper()
-	cmd := exec.Command("curl", "-sSfL",
-		"--connect-timeout", "30",
-		"--max-time", "600",
-		"--retry", "5",
-		"--retry-delay", "5",
-		"--retry-all-errors",
-		"-o", dest, url)
-	cmd.Stdout = GinkgoWriter
-	cmd.Stderr = GinkgoWriter
-	Expect(cmd.Run()).To(Succeed(), "failed to download %s", url)
+	var err error
+	for attempt := 1; attempt <= 6; attempt++ {
+		if attempt > 1 {
+			time.Sleep(5 * time.Second)
+		}
+		cmd := exec.Command("curl", "-sSfL",
+			"--connect-timeout", "30",
+			"-C", "-",
+			"--speed-limit", "1048576",
+			"--speed-time", "120",
+			"-o", dest, url)
+		cmd.Stdout = GinkgoWriter
+		cmd.Stderr = GinkgoWriter
+		if err = cmd.Run(); err == nil {
+			break
+		}
+	}
+	Expect(err).NotTo(HaveOccurred(), "failed to download %s", url)
 	fi, err := os.Stat(dest)
 	Expect(err).NotTo(HaveOccurred())
 	Expect(fi.Size()).To(BeNumerically(">", 1024), "downloaded file is suspiciously small")