fix(pool): handle worker start failures gracefully (#9337)

Author: AriPerkkio

packages/vitest/src/node/pools/pool.ts

@@ -1,3 +1,4 @@
+import type { Span } from '@opentelemetry/api'
 import type { ContextTestEnvironment } from '../../types/worker'
 import type { Logger } from '../logger'
 import type { StateManager } from '../state'
@@ -118,21 +119,27 @@ export class Pool {
           WORKER_START_TIMEOUT,
         )
 
-        await runner.start({ workerId: task.context.workerId }).finally(() => clearTimeout(id))
+        await runner.start({ workerId: task.context.workerId })
+          .catch(error =>
+            resolver.reject(
+              new Error(`[vitest-pool]: Failed to start ${task.worker} worker for test files ${formatFiles(task)}.`, { cause: error }),
+            ),
+          )
+          .finally(() => clearTimeout(id))
       }
 
-      const span = runner.startTracesSpan(`vitest.worker.${method}`)
-      // Start running the test in the worker
-      runner.request(method, task.context)
+      let span: Span | undefined
+
+      if (!resolver.isRejected) {
+        span = runner.startTracesSpan(`vitest.worker.${method}`)
+
+        // Start running the test in the worker
+        runner.request(method, task.context)
+      }
 
       await resolver.promise
-        .catch((error) => {
-          span.recordException(error)
-          throw error
-        })
-        .finally(() => {
-          span.end()
-        })
+        .catch(error => span?.recordException(error))
+        .finally(() => span?.end())
 
       const index = this.activeTasks.indexOf(activeTask)
       if (index !== -1) {
@@ -158,7 +165,7 @@ export class Pool {
         )
 
         this.exitPromises.push(
-          runner.stop()
+          runner.stop({ force: resolver.isRejected })
             .then(() => clearTimeout(id))
             .catch(error => this.logger.error(`[vitest-pool]: Failed to terminate ${task.worker} worker for test files ${formatFiles(task)}.`, error)),
         )
@@ -281,7 +288,17 @@ function withResolvers() {
     reject = rej
   })
 
-  return { resolve, reject, promise }
+  const resolver = {
+    promise,
+    resolve,
+    reject: (reason: unknown) => {
+      resolver.isRejected = true
+      reject(reason)
+    },
+    isRejected: false,
+  }
+
+  return resolver
 }
 
 function formatFiles(task: PoolTask) {

packages/vitest/src/node/pools/poolRunner.ts

@@ -15,6 +15,7 @@ enum RunnerState {
   IDLE = 'idle',
   STARTING = 'starting',
   STARTED = 'started',
+  START_FAILURE = 'start_failure',
   STOPPING = 'stopping',
   STOPPED = 'stopped',
 }
@@ -231,7 +232,7 @@ export class PoolRunner {
       this._state = RunnerState.STARTED
     }
     catch (error: any) {
-      this._state = RunnerState.IDLE
+      this._state = RunnerState.START_FAILURE
       startSpan?.recordException(error)
       throw error
     }

test/config/test/pool.test.ts

@@ -1,7 +1,7 @@
 import type { SerializedConfig } from 'vitest'
 import type { TestUserConfig } from 'vitest/node'
 import { normalize } from 'pathe'
-import { assert, describe, expect, test } from 'vitest'
+import { assert, describe, expect, test, vi } from 'vitest'
 import { runVitest, StableTestFileOrderSorter } from '../../test-utils'
 
 describe.each(['forks', 'threads', 'vmThreads', 'vmForks'])('%s', async (pool) => {
@@ -102,6 +102,33 @@ test('non-isolated happy-dom worker pool receives all testfiles at once', async
   `)
 })
 
+test('worker start failure should not hang', async () => {
+  const stop = vi.fn()
+
+  const { stdout, stderr } = await runVitest({
+    root: './fixtures/pool',
+    include: ['a.test.ts'],
+    pool: {
+      name: 'pool-with-crashing-workers',
+      // @ts-expect-error -- intentional
+      createPoolWorker: () => ({
+        start: () => Promise.reject(new Error('Mock')),
+        stop,
+        on() {},
+        off() {},
+        send() {},
+      }),
+    },
+  })
+
+  expect(stderr).toContain('Error: [vitest-pool]: Failed to start pool-with-crashing-workers worker for test files')
+  expect(stderr).toContain('a.test.ts')
+  expect(stderr).toContain('Caused by: Error: Mock')
+  expect(stdout).toContain('Errors  1 error')
+
+  expect(stop).toHaveBeenCalled()
+})
+
 async function getConfig<T = SerializedConfig>(options: Partial<TestUserConfig>, cliOptions: Partial<TestUserConfig> = {}): Promise<T> {
   let config: T | undefined