fix: apply configurable timeout to Ollama native client

src/api/providers/__tests__/native-ollama.spec.ts

@@ -3,14 +3,19 @@
 import { NativeOllamaHandler } from "../native-ollama"
 import { ApiHandlerOptions } from "../../../shared/api"
 import { getOllamaModels } from "../fetchers/ollama"
+import { getApiRequestTimeout } from "../utils/timeout-config"
 
 // Mock the ollama package
-const mockChat = vitest.fn()
+const { mockChat, MockOllama } = vitest.hoisted(() => {
+	const mockChat = vitest.fn()
+	const MockOllama = vitest.fn().mockImplementation(() => ({
+		chat: mockChat,
+	}))
+	return { mockChat, MockOllama }
+})
 vitest.mock("ollama", () => {
 	return {
-		Ollama: vitest.fn().mockImplementation(() => ({
-			chat: mockChat,
-		})),
+		Ollama: MockOllama,
 		Message: vitest.fn(),
 	}
 })
@@ -20,6 +25,13 @@ vitest.mock("../fetchers/ollama", () => ({
 	getOllamaModels: vitest.fn(),
 }))
 
+// Mock the timeout config
+vitest.mock("../utils/timeout-config", () => ({
+	getApiRequestTimeout: vitest.fn(),
+}))
+
+const mockGetApiRequestTimeout = vitest.mocked(getApiRequestTimeout)
+
 const mockGetOllamaModels = vitest.mocked(getOllamaModels)
 
 describe("NativeOllamaHandler", () => {
@@ -28,6 +40,9 @@ describe("NativeOllamaHandler", () => {
 	beforeEach(() => {
 		vitest.clearAllMocks()
 
+		// Default mock for timeout config (600s = 600000ms)
+		mockGetApiRequestTimeout.mockReturnValue(600_000)
+
 		// Default mock for getOllamaModels
 		mockGetOllamaModels.mockResolvedValue({
 			llama2: {
@@ -605,4 +620,64 @@ describe("NativeOllamaHandler", () => {
 			expect(firstEndIndex).toBeGreaterThan(lastPartialIndex)
 		})
 	})
+
+	describe("timeout configuration", () => {
+		it("should pass a custom fetch with timeout to the Ollama client", async () => {
+			mockGetApiRequestTimeout.mockReturnValue(900_000) // 900s
+
+			// Create a new handler to trigger ensureClient with the mocked timeout
+			const options: ApiHandlerOptions = {
+				apiModelId: "llama2",
+				ollamaModelId: "llama2",
+				ollamaBaseUrl: "http://localhost:11434",
+			}
+
+			const timeoutHandler = new NativeOllamaHandler(options)
+
+			mockChat.mockImplementation(async function* () {
+				yield { message: { content: "Response" } }
+			})
+
+			const stream = timeoutHandler.createMessage("System", [{ role: "user" as const, content: "Test" }])
+			for await (const _ of stream) {
+				// consume stream
+			}
+
+			// Verify Ollama constructor was called with a fetch option
+			expect(MockOllama).toHaveBeenCalledWith(
+				expect.objectContaining({
+					host: "http://localhost:11434",
+					fetch: expect.any(Function),
+				}),
+			)
+		})
+
+		it("should not pass custom fetch when timeout is undefined", async () => {
+			mockGetApiRequestTimeout.mockReturnValue(undefined)
+
+			const options: ApiHandlerOptions = {
+				apiModelId: "llama2",
+				ollamaModelId: "llama2",
+				ollamaBaseUrl: "http://localhost:11434",
+			}
+
+			const timeoutHandler = new NativeOllamaHandler(options)
+
+			mockChat.mockImplementation(async function* () {
+				yield { message: { content: "Response" } }
+			})
+
+			const stream = timeoutHandler.createMessage("System", [{ role: "user" as const, content: "Test" }])
+			for await (const _ of stream) {
+				// consume stream
+			}
+
+			// Verify Ollama constructor was called WITHOUT a fetch option
+			expect(MockOllama).toHaveBeenCalledWith(
+				expect.not.objectContaining({
+					fetch: expect.any(Function),
+				}),
+			)
+		})
+	})
 })

src/api/providers/native-ollama.ts

@@ -7,6 +7,7 @@ import { BaseProvider } from "./base-provider"
 import type { ApiHandlerOptions } from "../../shared/api"
 import { getOllamaModels } from "./fetchers/ollama"
 import { TagMatcher } from "../../utils/tag-matcher"
+import { getApiRequestTimeout } from "./utils/timeout-config"
 import type { SingleCompletionHandler, ApiHandlerCreateMessageMetadata } from "../index"
 
 interface OllamaChatOptions {
@@ -160,7 +161,20 @@ export class NativeOllamaHandler extends BaseProvider implements SingleCompletio
 			try {
 				const clientOptions: OllamaOptions = {
 					host: this.options.ollamaBaseUrl || "http://localhost:11434",
-					// Note: The ollama npm package handles timeouts internally
+				}
+
+				// Apply configurable timeout via custom fetch wrapper.
+				// The ollama npm package uses Node.js native fetch (Undici) which
+				// defaults to a 300s (5 minute) timeout. This respects the user's
+				// apiRequestTimeout setting (default 600s) to support slow inference.
+				const timeoutMs = getApiRequestTimeout()
+				if (timeoutMs) {
+					clientOptions.fetch = ((url: RequestInfo | URL, init?: RequestInit) => {
+						return fetch(url, {
+							...init,
+							signal: init?.signal ?? AbortSignal.timeout(timeoutMs),
+						})
+					}) as typeof fetch
 				}
 
 				// Add API key if provided (for Ollama cloud or authenticated instances)