File
Blob: tests/worker/project-do/watchdog-dispatch.test.ts
| 1 | import { runInDurableObject } from "cloudflare:test"; |
| 2 | import { env } from "cloudflare:workers"; |
| 3 | import { eq } from "drizzle-orm"; |
| 4 | import { describe, expect, it, vi } from "vitest"; |
| 5 | |
| 6 | import { DEFAULT_DISPATCH_MODE, DEFAULT_EXECUTION_RUNTIME, UnixTimestampMs } from "@/contracts"; |
| 7 | import { expectTrusted, PositiveInteger } from "@/worker/contracts"; |
| 8 | import { DISPATCH_RETRY_DELAYS_MS, HEARTBEAT_STALE_AFTER_MS } from "@/worker/durable/project-do/constants"; |
| 9 | import { createD1Db } from "@/worker/db/d1"; |
| 10 | import * as d1Schema from "@/worker/db/d1/schema"; |
| 11 | import { ProjectDO } from "@/worker/durable"; |
| 12 | import { |
| 13 | dispatchExecutableRun, |
| 14 | reconcileAcceptedRunD1Sync, |
| 15 | reconcileActiveRunWatchdog, |
| 16 | reconcileTerminalRunD1Sync, |
| 17 | } from "@/worker/durable/project-do/reconciliation"; |
| 18 | import { recoverWorkflowDispatchFailure } from "@/worker/durable/project-do/commands"; |
| 19 | import { |
| 20 | getDispatchRetryAt, |
| 21 | getSandboxCleanupRetryState, |
| 22 | setHeartbeatAt, |
| 23 | } from "@/worker/durable/project-do/sidecar-state"; |
| 24 | import type { ProjectDoContext } from "@/worker/durable/project-do/types"; |
| 25 | |
| 26 | import { |
| 27 | acceptManualRunWithoutAlarm, |
| 28 | claimRunWorkWithoutAlarm, |
| 29 | createTestProjectDoContext, |
| 30 | expectAcceptedManualRun, |
| 31 | } from "../../helpers/project-do"; |
| 32 | import { registerWorkerRuntimeHooks } from "../../helpers/worker-hooks"; |
| 33 | import { readProjectDoRows, seedProject, seedUser } from "../../helpers/runtime"; |
| 34 | |
| 35 | describe("ProjectDO watchdog and dispatch recovery", () => { |
| 36 | registerWorkerRuntimeHooks(); |
| 37 | |
| 38 | it("fails stale active runs as runner_lost during watchdog recovery", async () => { |
| 39 | const baseTime = 1_710_000_000_000; |
| 40 | const nowSpy = vi.spyOn(Date, "now").mockImplementation(() => baseTime); |
| 41 | |
| 42 | try { |
| 43 | const user = await seedUser({ |
| 44 | email: "watchdog@example.com", |
| 45 | slug: "watchdog-user", |
| 46 | }); |
| 47 | const project = await seedProject(user, { |
| 48 | projectSlug: "watchdog-project", |
| 49 | }); |
| 50 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 51 | |
| 52 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 53 | projectId: project.id, |
| 54 | triggeredByUserId: user.id, |
| 55 | branch: project.defaultBranch, |
| 56 | }); |
| 57 | |
| 58 | const claim = await claimRunWorkWithoutAlarm(projectStub, { |
| 59 | projectId: project.id, |
| 60 | runId: accepted.runId, |
| 61 | }); |
| 62 | expect(claim.kind).toBe("execute"); |
| 63 | |
| 64 | nowSpy.mockImplementation(() => baseTime + HEARTBEAT_STALE_AFTER_MS + 1); |
| 65 | await runInDurableObject(env.PROJECT_DO.getByName(project.id), async (instance: ProjectDO) => { |
| 66 | await instance.alarm(); |
| 67 | }); |
| 68 | |
| 69 | const rows = await readProjectDoRows(project.id); |
| 70 | expect(rows.state?.activeRunId).toBeNull(); |
| 71 | expect(rows.runs[0]?.status).toBe("failed"); |
| 72 | expect(rows.runs[0]?.lastError).toBe("runner_lost"); |
| 73 | |
| 74 | const runMeta = await env.RUN_DO.getByName(accepted.runId).getRunSummary(accepted.runId); |
| 75 | expect(runMeta?.status).toBe("failed"); |
| 76 | expect(runMeta?.errorMessage).toBe("runner_lost"); |
| 77 | expect(runMeta?.startedAt).toBeNull(); |
| 78 | |
| 79 | const db = createD1Db(env.DB); |
| 80 | const d1Row = await db.select().from(d1Schema.runIndex).where(eq(d1Schema.runIndex.id, accepted.runId)).limit(1); |
| 81 | expect(d1Row[0]?.status).toBe("failed"); |
| 82 | expect(d1Row[0]?.startedAt).toBeNull(); |
| 83 | } finally { |
| 84 | nowSpy.mockRestore(); |
| 85 | } |
| 86 | }); |
| 87 | |
| 88 | it("repairs the active step before failing a stale active run during watchdog recovery", async () => { |
| 89 | const baseTime = 1_712_000_000_000; |
| 90 | const startedAt = expectTrusted(UnixTimestampMs, baseTime, "UnixTimestampMs"); |
| 91 | const user = await seedUser({ |
| 92 | email: "watchdog-step-repair@example.com", |
| 93 | slug: "watchdog-step-repair-user", |
| 94 | }); |
| 95 | const project = await seedProject(user, { |
| 96 | projectSlug: "watchdog-step-repair-project", |
| 97 | }); |
| 98 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 99 | |
| 100 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 101 | projectId: project.id, |
| 102 | triggeredByUserId: user.id, |
| 103 | branch: project.defaultBranch, |
| 104 | }); |
| 105 | |
| 106 | const claim = await claimRunWorkWithoutAlarm(projectStub, { |
| 107 | projectId: project.id, |
| 108 | runId: accepted.runId, |
| 109 | }); |
| 110 | expect(claim.kind).toBe("execute"); |
| 111 | |
| 112 | const stepPosition = expectTrusted(PositiveInteger, 1, "PositiveInteger"); |
| 113 | const runStub = env.RUN_DO.getByName(accepted.runId); |
| 114 | await runStub.replaceSteps({ |
| 115 | runId: accepted.runId, |
| 116 | steps: [ |
| 117 | { |
| 118 | position: stepPosition, |
| 119 | name: "Install", |
| 120 | command: "npm ci", |
| 121 | }, |
| 122 | ], |
| 123 | }); |
| 124 | await runStub.updateRunState({ |
| 125 | runId: accepted.runId, |
| 126 | status: "starting", |
| 127 | currentStep: null, |
| 128 | startedAt, |
| 129 | finishedAt: null, |
| 130 | exitCode: null, |
| 131 | errorMessage: null, |
| 132 | }); |
| 133 | await runStub.updateStepState({ |
| 134 | runId: accepted.runId, |
| 135 | position: stepPosition, |
| 136 | status: "running", |
| 137 | startedAt, |
| 138 | finishedAt: null, |
| 139 | exitCode: null, |
| 140 | }); |
| 141 | await runStub.updateRunState({ |
| 142 | runId: accepted.runId, |
| 143 | status: "running", |
| 144 | currentStep: stepPosition, |
| 145 | startedAt, |
| 146 | finishedAt: null, |
| 147 | exitCode: null, |
| 148 | errorMessage: null, |
| 149 | }); |
| 150 | |
| 151 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 152 | const context = createTestProjectDoContext(instance); |
| 153 | await setHeartbeatAt(context, accepted.runId, baseTime); |
| 154 | await expect(reconcileActiveRunWatchdog(context, project.id)).resolves.toBe(accepted.runId); |
| 155 | }); |
| 156 | |
| 157 | const detail = await runStub.getRunDetail(accepted.runId); |
| 158 | expect(detail.meta?.status).toBe("failed"); |
| 159 | expect(detail.meta?.currentStep).toBeNull(); |
| 160 | expect(detail.meta?.errorMessage).toBe("runner_lost"); |
| 161 | expect(detail.steps[0]?.status).toBe("failed"); |
| 162 | expect(detail.steps[0]?.finishedAt).not.toBeNull(); |
| 163 | }); |
| 164 | |
| 165 | it("persists sandbox cleanup retry state when watchdog cleanup fails", async () => { |
| 166 | const baseTime = 1_715_000_000_000; |
| 167 | const nowSpy = vi.spyOn(Date, "now").mockImplementation(() => baseTime); |
| 168 | |
| 169 | try { |
| 170 | const user = await seedUser({ |
| 171 | email: "watchdog-cleanup@example.com", |
| 172 | slug: "watchdog-cleanup-user", |
| 173 | }); |
| 174 | const project = await seedProject(user, { |
| 175 | projectSlug: "watchdog-cleanup-project", |
| 176 | }); |
| 177 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 178 | |
| 179 | const accepted = expectAcceptedManualRun( |
| 180 | await projectStub.acceptManualRun({ |
| 181 | projectId: project.id, |
| 182 | triggeredByUserId: user.id, |
| 183 | branch: project.defaultBranch, |
| 184 | }), |
| 185 | ); |
| 186 | |
| 187 | const claim = await projectStub.claimRunWork({ |
| 188 | projectId: project.id, |
| 189 | runId: accepted.runId, |
| 190 | }); |
| 191 | expect(claim.kind).toBe("execute"); |
| 192 | |
| 193 | nowSpy.mockImplementation(() => baseTime + HEARTBEAT_STALE_AFTER_MS + 1); |
| 194 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 195 | const baseContext = createTestProjectDoContext(instance); |
| 196 | const context: ProjectDoContext = { |
| 197 | ...baseContext, |
| 198 | env: Object.assign(Object.create(baseContext.env), { |
| 199 | Sandbox: Object.assign(Object.create(baseContext.env.Sandbox), { |
| 200 | getByName: (sandboxId: string) => { |
| 201 | const stub = baseContext.env.Sandbox.getByName(sandboxId); |
| 202 | if (sandboxId !== accepted.runId) { |
| 203 | return stub; |
| 204 | } |
| 205 | |
| 206 | return Object.assign(Object.create(stub), { |
| 207 | setKeepAlive: vi.fn(async () => {}), |
| 208 | destroy: vi.fn(async () => { |
| 209 | throw new Error("sandbox still reachable"); |
| 210 | }), |
| 211 | }); |
| 212 | }, |
| 213 | }), |
| 214 | }) as Env, |
| 215 | }; |
| 216 | |
| 217 | await expect(reconcileActiveRunWatchdog(context, project.id)).resolves.toBe(accepted.runId); |
| 218 | |
| 219 | const retryState = await getSandboxCleanupRetryState(baseContext, accepted.runId); |
| 220 | expect(retryState?.attempt).toBe(1); |
| 221 | expect(retryState?.nextAt ?? 0).toBeGreaterThan(Date.now()); |
| 222 | }); |
| 223 | } finally { |
| 224 | nowSpy.mockRestore(); |
| 225 | } |
| 226 | }); |
| 227 | |
| 228 | it("marks a run failed after repeated dispatch enqueue failures", async () => { |
| 229 | const baseTime = 1_720_000_000_000; |
| 230 | const nowSpy = vi.spyOn(Date, "now").mockImplementation(() => baseTime); |
| 231 | |
| 232 | try { |
| 233 | const user = await seedUser({ |
| 234 | email: "dispatch@example.com", |
| 235 | slug: "dispatch-user", |
| 236 | }); |
| 237 | const project = await seedProject(user, { |
| 238 | projectSlug: "dispatch-project", |
| 239 | dispatchMode: "queue", |
| 240 | }); |
| 241 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 242 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 243 | projectId: project.id, |
| 244 | triggeredByUserId: user.id, |
| 245 | branch: project.defaultBranch, |
| 246 | }); |
| 247 | |
| 248 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 249 | const baseContext = createTestProjectDoContext(instance); |
| 250 | const context: ProjectDoContext = { |
| 251 | ...baseContext, |
| 252 | env: Object.assign(Object.create(baseContext.env), { |
| 253 | RUN_QUEUE: { |
| 254 | send: async () => { |
| 255 | throw new Error("queue unavailable"); |
| 256 | }, |
| 257 | }, |
| 258 | }) as Env, |
| 259 | }; |
| 260 | await baseContext.ctx.storage.deleteAlarm(); |
| 261 | |
| 262 | await reconcileAcceptedRunD1Sync(context, project.id); |
| 263 | |
| 264 | for (let attempt = 0; attempt <= DISPATCH_RETRY_DELAYS_MS.length; attempt += 1) { |
| 265 | await dispatchExecutableRun(context, project.id); |
| 266 | if (attempt < DISPATCH_RETRY_DELAYS_MS.length) { |
| 267 | nowSpy.mockImplementation( |
| 268 | () => |
| 269 | baseTime + DISPATCH_RETRY_DELAYS_MS.slice(0, attempt + 1).reduce((sum, delay) => sum + delay + 1, 0), |
| 270 | ); |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | await reconcileTerminalRunD1Sync(context, project.id); |
| 275 | }); |
| 276 | |
| 277 | const rows = await readProjectDoRows(project.id); |
| 278 | expect(rows.runs[0]?.status).toBe("failed"); |
| 279 | expect(rows.runs[0]?.dispatchStatus).toBe("terminal"); |
| 280 | expect(rows.runs[0]?.lastError).toBe("dispatch_failed"); |
| 281 | |
| 282 | const runMeta = await env.RUN_DO.getByName(accepted.runId).getRunSummary(accepted.runId); |
| 283 | expect(runMeta?.status).toBe("failed"); |
| 284 | expect(runMeta?.errorMessage).toBe("dispatch_failed"); |
| 285 | expect(runMeta?.startedAt).toBeNull(); |
| 286 | |
| 287 | const db = createD1Db(env.DB); |
| 288 | const d1Row = await db.select().from(d1Schema.runIndex).where(eq(d1Schema.runIndex.id, accepted.runId)).limit(1); |
| 289 | expect(d1Row[0]?.status).toBe("failed"); |
| 290 | expect(d1Row[0]?.startedAt).toBeNull(); |
| 291 | } finally { |
| 292 | nowSpy.mockRestore(); |
| 293 | } |
| 294 | }); |
| 295 | |
| 296 | it("starts a Workflows instance when the run dispatch mode is workflows", async () => { |
| 297 | const user = await seedUser({ |
| 298 | email: "workflow-dispatch@example.com", |
| 299 | slug: "workflow-dispatch-user", |
| 300 | }); |
| 301 | const project = await seedProject(user, { |
| 302 | projectSlug: "workflow-dispatch-project", |
| 303 | dispatchMode: "workflows", |
| 304 | }); |
| 305 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 306 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 307 | projectId: project.id, |
| 308 | triggeredByUserId: user.id, |
| 309 | branch: project.defaultBranch, |
| 310 | }); |
| 311 | |
| 312 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 313 | const baseContext = createTestProjectDoContext(instance); |
| 314 | const createBatch = vi.fn(async () => [{ id: accepted.runId }]); |
| 315 | const context: ProjectDoContext = { |
| 316 | ...baseContext, |
| 317 | env: Object.assign(Object.create(baseContext.env), { |
| 318 | RUN_WORKFLOWS: { |
| 319 | createBatch, |
| 320 | get: vi.fn(), |
| 321 | }, |
| 322 | }) as Env, |
| 323 | }; |
| 324 | await baseContext.ctx.storage.deleteAlarm(); |
| 325 | |
| 326 | await dispatchExecutableRun(context, project.id); |
| 327 | |
| 328 | expect(createBatch).toHaveBeenCalledWith([ |
| 329 | expect.objectContaining({ |
| 330 | id: accepted.runId, |
| 331 | params: expect.objectContaining({ |
| 332 | runId: accepted.runId, |
| 333 | projectId: project.id, |
| 334 | dispatchMode: "workflows", |
| 335 | executionRuntime: DEFAULT_EXECUTION_RUNTIME, |
| 336 | }), |
| 337 | }), |
| 338 | ]); |
| 339 | }); |
| 340 | |
| 341 | const rows = await readProjectDoRows(project.id); |
| 342 | expect(rows.runs[0]?.dispatchStatus).toBe("queued"); |
| 343 | }); |
| 344 | |
| 345 | it.each([ |
| 346 | { |
| 347 | workflowStatus: "errored" as const, |
| 348 | expectedError: "reached terminal status errored", |
| 349 | }, |
| 350 | { |
| 351 | workflowStatus: "terminated" as const, |
| 352 | expectedError: "reached terminal status terminated", |
| 353 | }, |
| 354 | ])( |
| 355 | "rearms a queued workflow dispatch when the retained instance is $workflowStatus", |
| 356 | async ({ workflowStatus, expectedError }) => { |
| 357 | const user = await seedUser({ |
| 358 | email: `workflow-${workflowStatus}-queued@example.com`, |
| 359 | slug: `workflow-${workflowStatus}-queued-user`, |
| 360 | }); |
| 361 | const project = await seedProject(user, { |
| 362 | projectSlug: `workflow-${workflowStatus}-queued-project`, |
| 363 | dispatchMode: "workflows", |
| 364 | }); |
| 365 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 366 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 367 | projectId: project.id, |
| 368 | triggeredByUserId: user.id, |
| 369 | branch: project.defaultBranch, |
| 370 | }); |
| 371 | |
| 372 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 373 | const baseContext = createTestProjectDoContext(instance); |
| 374 | const status = vi.fn(async () => ({ |
| 375 | status: workflowStatus, |
| 376 | })); |
| 377 | const context: ProjectDoContext = { |
| 378 | ...baseContext, |
| 379 | env: Object.assign(Object.create(baseContext.env), { |
| 380 | RUN_WORKFLOWS: { |
| 381 | createBatch: vi.fn(async () => [{ id: accepted.runId }]), |
| 382 | get: vi.fn(async () => ({ |
| 383 | status, |
| 384 | restart: vi.fn(async () => {}), |
| 385 | })), |
| 386 | }, |
| 387 | }) as Env, |
| 388 | }; |
| 389 | await baseContext.ctx.storage.deleteAlarm(); |
| 390 | |
| 391 | await dispatchExecutableRun(context, project.id); |
| 392 | await expect(dispatchExecutableRun(context, project.id)).resolves.toBeNull(); |
| 393 | |
| 394 | const retryAt = await getDispatchRetryAt(baseContext, accepted.runId); |
| 395 | expect(retryAt ?? 0).toBeGreaterThan(Date.now()); |
| 396 | }); |
| 397 | |
| 398 | const rows = await readProjectDoRows(project.id); |
| 399 | expect(rows.runs[0]?.status).toBe("executable"); |
| 400 | expect(rows.runs[0]?.dispatchStatus).toBe("pending"); |
| 401 | expect(rows.runs[0]?.dispatchAttempts).toBe(1); |
| 402 | expect(rows.runs[0]?.lastError).toContain(expectedError); |
| 403 | }, |
| 404 | ); |
| 405 | |
| 406 | it("rearms a queued workflow dispatch when workflow status inspection fails", async () => { |
| 407 | const user = await seedUser({ |
| 408 | email: "workflow-queued-status-throw@example.com", |
| 409 | slug: "workflow-queued-status-throw-user", |
| 410 | }); |
| 411 | const project = await seedProject(user, { |
| 412 | projectSlug: "workflow-queued-status-throw-project", |
| 413 | dispatchMode: "workflows", |
| 414 | }); |
| 415 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 416 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 417 | projectId: project.id, |
| 418 | triggeredByUserId: user.id, |
| 419 | branch: project.defaultBranch, |
| 420 | }); |
| 421 | |
| 422 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 423 | const baseContext = createTestProjectDoContext(instance); |
| 424 | const context: ProjectDoContext = { |
| 425 | ...baseContext, |
| 426 | env: Object.assign(Object.create(baseContext.env), { |
| 427 | RUN_WORKFLOWS: { |
| 428 | createBatch: vi.fn(async () => [{ id: accepted.runId }]), |
| 429 | get: vi.fn(async () => ({ |
| 430 | status: vi.fn(async () => { |
| 431 | throw new Error("workflow status unavailable"); |
| 432 | }), |
| 433 | restart: vi.fn(async () => {}), |
| 434 | })), |
| 435 | }, |
| 436 | }) as Env, |
| 437 | }; |
| 438 | await baseContext.ctx.storage.deleteAlarm(); |
| 439 | |
| 440 | await dispatchExecutableRun(context, project.id); |
| 441 | await expect(dispatchExecutableRun(context, project.id)).resolves.toBeNull(); |
| 442 | |
| 443 | const retryAt = await getDispatchRetryAt(baseContext, accepted.runId); |
| 444 | expect(retryAt ?? 0).toBeGreaterThan(Date.now()); |
| 445 | }); |
| 446 | |
| 447 | const rows = await readProjectDoRows(project.id); |
| 448 | expect(rows.runs[0]?.status).toBe("executable"); |
| 449 | expect(rows.runs[0]?.dispatchStatus).toBe("pending"); |
| 450 | expect(rows.runs[0]?.dispatchAttempts).toBe(1); |
| 451 | expect(rows.runs[0]?.lastError).toContain("workflow status unavailable"); |
| 452 | }); |
| 453 | |
| 454 | it("preserves a workflow dispatch retry when the workflow rearms itself before dispatch returns", async () => { |
| 455 | const baseTime = 1_730_500_000_000; |
| 456 | const nowSpy = vi.spyOn(Date, "now").mockImplementation(() => baseTime); |
| 457 | |
| 458 | try { |
| 459 | const user = await seedUser({ |
| 460 | email: "workflow-fast-failure@example.com", |
| 461 | slug: "workflow-fast-failure-user", |
| 462 | }); |
| 463 | const project = await seedProject(user, { |
| 464 | projectSlug: "workflow-fast-failure-project", |
| 465 | dispatchMode: "workflows", |
| 466 | }); |
| 467 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 468 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 469 | projectId: project.id, |
| 470 | triggeredByUserId: user.id, |
| 471 | branch: project.defaultBranch, |
| 472 | }); |
| 473 | |
| 474 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 475 | const baseContext = createTestProjectDoContext(instance); |
| 476 | let context: ProjectDoContext; |
| 477 | context = { |
| 478 | ...baseContext, |
| 479 | env: Object.assign(Object.create(baseContext.env), { |
| 480 | RUN_WORKFLOWS: { |
| 481 | createBatch: vi.fn(async () => { |
| 482 | const recovery = await recoverWorkflowDispatchFailure(baseContext, { |
| 483 | projectId: project.id, |
| 484 | runId: accepted.runId, |
| 485 | errorMessage: "claim exploded", |
| 486 | }); |
| 487 | expect(recovery).toEqual({ |
| 488 | kind: "rearmed", |
| 489 | }); |
| 490 | return [{ id: accepted.runId }]; |
| 491 | }), |
| 492 | get: vi.fn(), |
| 493 | }, |
| 494 | }) as Env, |
| 495 | }; |
| 496 | await baseContext.ctx.storage.deleteAlarm(); |
| 497 | |
| 498 | await dispatchExecutableRun(context, project.id); |
| 499 | |
| 500 | const retryAt = await getDispatchRetryAt(baseContext, accepted.runId); |
| 501 | expect(retryAt ?? 0).toBeGreaterThan(Date.now()); |
| 502 | }); |
| 503 | |
| 504 | const rows = await readProjectDoRows(project.id); |
| 505 | expect(rows.runs[0]?.status).toBe("executable"); |
| 506 | expect(rows.runs[0]?.dispatchStatus).toBe("pending"); |
| 507 | expect(rows.runs[0]?.dispatchAttempts).toBe(1); |
| 508 | expect(rows.runs[0]?.lastError).toBe("claim exploded"); |
| 509 | } finally { |
| 510 | nowSpy.mockRestore(); |
| 511 | } |
| 512 | }); |
| 513 | |
| 514 | it("restarts an existing Workflows instance after a rearmed pre-active failure", async () => { |
| 515 | const baseTime = 1_730_000_000_000; |
| 516 | const nowSpy = vi.spyOn(Date, "now").mockImplementation(() => baseTime); |
| 517 | try { |
| 518 | const user = await seedUser({ |
| 519 | email: "workflow-restart@example.com", |
| 520 | slug: "workflow-restart-user", |
| 521 | }); |
| 522 | const project = await seedProject(user, { |
| 523 | projectSlug: "workflow-restart-project", |
| 524 | dispatchMode: "workflows", |
| 525 | }); |
| 526 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 527 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 528 | projectId: project.id, |
| 529 | triggeredByUserId: user.id, |
| 530 | branch: project.defaultBranch, |
| 531 | }); |
| 532 | |
| 533 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 534 | const baseContext = createTestProjectDoContext(instance); |
| 535 | const restart = vi.fn(async () => {}); |
| 536 | const context: ProjectDoContext = { |
| 537 | ...baseContext, |
| 538 | env: Object.assign(Object.create(baseContext.env), { |
| 539 | RUN_WORKFLOWS: { |
| 540 | createBatch: vi.fn(async () => [{ id: accepted.runId }]), |
| 541 | get: vi.fn(async () => ({ |
| 542 | restart, |
| 543 | })), |
| 544 | }, |
| 545 | }) as Env, |
| 546 | }; |
| 547 | await baseContext.ctx.storage.deleteAlarm(); |
| 548 | |
| 549 | await dispatchExecutableRun(context, project.id); |
| 550 | const recovery = await recoverWorkflowDispatchFailure(context, { |
| 551 | projectId: project.id, |
| 552 | runId: accepted.runId, |
| 553 | errorMessage: "claim exploded", |
| 554 | }); |
| 555 | expect(recovery).toEqual({ |
| 556 | kind: "rearmed", |
| 557 | }); |
| 558 | nowSpy.mockImplementation(() => baseTime + DISPATCH_RETRY_DELAYS_MS[0] + 1); |
| 559 | |
| 560 | await dispatchExecutableRun( |
| 561 | { |
| 562 | ...context, |
| 563 | env: Object.assign(Object.create(context.env), { |
| 564 | RUN_WORKFLOWS: { |
| 565 | createBatch: vi.fn(async () => []), |
| 566 | get: vi.fn(async () => ({ |
| 567 | status: vi.fn(async () => ({ |
| 568 | status: "complete" as const, |
| 569 | })), |
| 570 | restart, |
| 571 | })), |
| 572 | }, |
| 573 | }) as Env, |
| 574 | }, |
| 575 | project.id, |
| 576 | ); |
| 577 | |
| 578 | expect(restart).toHaveBeenCalledTimes(1); |
| 579 | }); |
| 580 | } finally { |
| 581 | nowSpy.mockRestore(); |
| 582 | } |
| 583 | }); |
| 584 | |
| 585 | it("does not restart an already-running Workflows instance when dispatch is replayed", async () => { |
| 586 | const user = await seedUser({ |
| 587 | email: "workflow-running@example.com", |
| 588 | slug: "workflow-running-user", |
| 589 | }); |
| 590 | const project = await seedProject(user, { |
| 591 | projectSlug: "workflow-running-project", |
| 592 | dispatchMode: "workflows", |
| 593 | }); |
| 594 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 595 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 596 | projectId: project.id, |
| 597 | triggeredByUserId: user.id, |
| 598 | branch: project.defaultBranch, |
| 599 | }); |
| 600 | |
| 601 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 602 | const baseContext = createTestProjectDoContext(instance); |
| 603 | const restart = vi.fn(async () => {}); |
| 604 | const status = vi.fn(async () => ({ |
| 605 | status: "running" as const, |
| 606 | })); |
| 607 | const context: ProjectDoContext = { |
| 608 | ...baseContext, |
| 609 | env: Object.assign(Object.create(baseContext.env), { |
| 610 | RUN_WORKFLOWS: { |
| 611 | createBatch: vi.fn(async () => []), |
| 612 | get: vi.fn(async () => ({ |
| 613 | status, |
| 614 | restart, |
| 615 | })), |
| 616 | }, |
| 617 | }) as Env, |
| 618 | }; |
| 619 | await baseContext.ctx.storage.deleteAlarm(); |
| 620 | |
| 621 | await dispatchExecutableRun(context, project.id); |
| 622 | |
| 623 | expect(status).toHaveBeenCalledTimes(1); |
| 624 | expect(restart).not.toHaveBeenCalled(); |
| 625 | }); |
| 626 | |
| 627 | const rows = await readProjectDoRows(project.id); |
| 628 | expect(rows.runs[0]?.dispatchStatus).toBe("queued"); |
| 629 | }); |
| 630 | |
| 631 | it.each(["paused", "waitingForPause"] as const)( |
| 632 | "does not restart a retained Workflows instance when dispatch is replayed in %s state", |
| 633 | async (workflowStatus) => { |
| 634 | const user = await seedUser({ |
| 635 | email: `workflow-${workflowStatus.toLowerCase()}@example.com`, |
| 636 | slug: `workflow-${workflowStatus.toLowerCase()}-user`, |
| 637 | }); |
| 638 | const project = await seedProject(user, { |
| 639 | projectSlug: `workflow-${workflowStatus.toLowerCase()}-project`, |
| 640 | dispatchMode: "workflows", |
| 641 | }); |
| 642 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 643 | const accepted = await acceptManualRunWithoutAlarm(projectStub, { |
| 644 | projectId: project.id, |
| 645 | triggeredByUserId: user.id, |
| 646 | branch: project.defaultBranch, |
| 647 | }); |
| 648 | |
| 649 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 650 | const baseContext = createTestProjectDoContext(instance); |
| 651 | const restart = vi.fn(async () => {}); |
| 652 | const status = vi.fn(async () => ({ |
| 653 | status: workflowStatus, |
| 654 | })); |
| 655 | const context: ProjectDoContext = { |
| 656 | ...baseContext, |
| 657 | env: Object.assign(Object.create(baseContext.env), { |
| 658 | RUN_WORKFLOWS: { |
| 659 | createBatch: vi.fn(async () => []), |
| 660 | get: vi.fn(async () => ({ |
| 661 | status, |
| 662 | restart, |
| 663 | })), |
| 664 | }, |
| 665 | }) as Env, |
| 666 | }; |
| 667 | await baseContext.ctx.storage.deleteAlarm(); |
| 668 | |
| 669 | await dispatchExecutableRun(context, project.id); |
| 670 | |
| 671 | expect(status).toHaveBeenCalledTimes(1); |
| 672 | expect(restart).not.toHaveBeenCalled(); |
| 673 | }); |
| 674 | |
| 675 | const rows = await readProjectDoRows(project.id); |
| 676 | expect(rows.runs[0]?.dispatchStatus).toBe("queued"); |
| 677 | }, |
| 678 | ); |
| 679 | |
| 680 | it("keeps workflow dispatch pending when an existing instance has an unsupported status", async () => { |
| 681 | const user = await seedUser({ |
| 682 | email: "workflow-unknown@example.com", |
| 683 | slug: "workflow-unknown-user", |
| 684 | }); |
| 685 | const project = await seedProject(user, { |
| 686 | projectSlug: "workflow-unknown-project", |
| 687 | dispatchMode: "workflows", |
| 688 | }); |
| 689 | const projectStub = env.PROJECT_DO.getByName(project.id); |
| 690 | await acceptManualRunWithoutAlarm(projectStub, { |
| 691 | projectId: project.id, |
| 692 | triggeredByUserId: user.id, |
| 693 | branch: project.defaultBranch, |
| 694 | }); |
| 695 | |
| 696 | await runInDurableObject(projectStub, async (instance: ProjectDO) => { |
| 697 | const baseContext = createTestProjectDoContext(instance); |
| 698 | const context: ProjectDoContext = { |
| 699 | ...baseContext, |
| 700 | env: Object.assign(Object.create(baseContext.env), { |
| 701 | RUN_WORKFLOWS: { |
| 702 | createBatch: vi.fn(async () => []), |
| 703 | get: vi.fn(async () => ({ |
| 704 | status: vi.fn(async () => ({ |
| 705 | status: "unknown" as const, |
| 706 | })), |
| 707 | restart: vi.fn(async () => {}), |
| 708 | })), |
| 709 | }, |
| 710 | }) as Env, |
| 711 | }; |
| 712 | await baseContext.ctx.storage.deleteAlarm(); |
| 713 | |
| 714 | await expect(dispatchExecutableRun(context, project.id)).resolves.toBeNull(); |
| 715 | }); |
| 716 | |
| 717 | const rows = await readProjectDoRows(project.id); |
| 718 | expect(rows.runs[0]?.dispatchStatus).toBe("pending"); |
| 719 | expect(rows.runs[0]?.dispatchAttempts).toBe(1); |
| 720 | expect(rows.runs[0]?.lastError).toContain("unsupported status unknown"); |
| 721 | }); |
| 722 | }); |