test_fetch_mpack_prebuild.py
file-level
1
files
1
commits
0
hotspots
0
π§ dead
0
π₯ blast risk
| 1 | """TDD β fetch.mpack.prebuild job handler and wire_fetch_mpack cache (issue #92 Phases 2β5). |
| 2 | |
| 3 | Test IDs: |
| 4 | FMC_07 Unit: mock wire_fetch_mpack, confirm cache rows written for each tip, |
| 5 | confirm existing fresh entries are skipped |
| 6 | FMC_08 Integration: insert a job row, run the handler, verify |
| 7 | MusehubFetchMPackCache row exists with correct repo_id/tip/mpack_id |
| 8 | FMC_13 Unit: cache hit returns presigned URL without entering blob-load path |
| 9 | FMC_14 Unit: cache miss builds and writes cache row |
| 10 | FMC_18 Integration: enqueue_push_intel inserts fetch.mpack.prebuild with branch tips |
| 11 | FMC_20 Unit: gc_fetch_mpack_cache deletes expired rows + R2 objects; fresh rows untouched |
| 12 | """ |
| 13 | from __future__ import annotations |
| 14 | |
| 15 | import hashlib |
| 16 | from datetime import datetime, timedelta, timezone |
| 17 | from unittest.mock import AsyncMock, patch |
| 18 | |
| 19 | import pytest |
| 20 | from sqlalchemy import select |
| 21 | from sqlalchemy.ext.asyncio import AsyncSession |
| 22 | |
| 23 | from musehub.core.genesis import compute_job_id |
| 24 | from musehub.db.musehub_jobs_models import MusehubBackgroundJob |
| 25 | from musehub.db.musehub_repo_models import MusehubFetchMPackCache |
| 26 | from musehub.services.musehub_gc import gc_fetch_mpack_cache |
| 27 | from musehub.services.musehub_jobs import enqueue_push_intel |
| 28 | from musehub.services.musehub_wire_fetch import process_fetch_mpack_prebuild_job, wire_fetch_mpack |
| 29 | from musehub.services.musehub_wire_shared import MPackNotReadyError |
| 30 | from tests.factories import create_branch, create_repo |
| 31 | |
| 32 | |
| 33 | def _now() -> datetime: |
| 34 | return datetime.now(tz=timezone.utc) |
| 35 | |
| 36 | |
| 37 | def _fake_commit_id(seed: str) -> str: |
| 38 | return "sha256:" + hashlib.sha256(seed.encode()).hexdigest() |
| 39 | |
| 40 | |
| 41 | def _fake_mpack_id(seed: str) -> str: |
| 42 | return "sha256:" + hashlib.sha256(f"mpack-{seed}".encode()).hexdigest() |
| 43 | |
| 44 | |
| 45 | async def _insert_job( |
| 46 | session: AsyncSession, |
| 47 | repo_id: str, |
| 48 | tip_commit_ids: list[str], |
| 49 | ) -> str: |
| 50 | now = _now() |
| 51 | job_id = compute_job_id(repo_id, "fetch.mpack.prebuild", now.isoformat()) |
| 52 | session.add(MusehubBackgroundJob( |
| 53 | job_id=job_id, |
| 54 | repo_id=repo_id, |
| 55 | job_type="fetch.mpack.prebuild", |
| 56 | payload={"tip_commit_ids": tip_commit_ids}, |
| 57 | status="pending", |
| 58 | created_at=now, |
| 59 | attempt=0, |
| 60 | )) |
| 61 | await session.flush() |
| 62 | return job_id |
| 63 | |
| 64 | |
| 65 | # ββ FMC_07 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 66 | |
| 67 | @pytest.mark.tier2 |
| 68 | async def test_fmc_07_builds_uncached_tips_in_one_combined_mpack(db_session: AsyncSession) -> None: |
| 69 | """FMC_07a: the handler builds all uncached tips in a SINGLE combined wire_fetch_mpack |
| 70 | call (want=[all uncached tips]) β not one call per tip. The per-tip cache rows are |
| 71 | written inside wire_fetch_mpack (covered by FMC_14), so it is mocked here and we assert |
| 72 | the handler's orchestration + counts only.""" |
| 73 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 74 | tip_a = _fake_commit_id("tip-a") |
| 75 | tip_b = _fake_commit_id("tip-b") |
| 76 | combined_mpack = _fake_mpack_id("combined") |
| 77 | |
| 78 | # The handler reads live branch tips from MusehubBranch (Phase 1 fix) β |
| 79 | # create branches so the handler sees tip_a and tip_b as the live tip set. |
| 80 | await create_branch(db_session, repo.repo_id, name="main", head_commit_id=tip_a) |
| 81 | await create_branch(db_session, repo.repo_id, name="dev", head_commit_id=tip_b) |
| 82 | job_id = await _insert_job(db_session, repo.repo_id, [tip_a, tip_b]) |
| 83 | await db_session.commit() |
| 84 | |
| 85 | with patch( |
| 86 | "musehub.services.musehub_wire_fetch.wire_fetch_mpack", |
| 87 | new_callable=AsyncMock, |
| 88 | return_value={"mpack_id": combined_mpack, "mpack_url": "https://r2.example/c", "commit_count": 2, "blob_count": 5}, |
| 89 | ) as mock_build: |
| 90 | result = await process_fetch_mpack_prebuild_job(db_session, job_id) |
| 91 | await db_session.commit() |
| 92 | |
| 93 | # One combined build covering both uncached tips β not one call per tip. |
| 94 | assert mock_build.call_count == 1 |
| 95 | assert set(mock_build.call_args.kwargs["want"]) == {tip_a, tip_b} |
| 96 | assert result["tips_requested"] == 2 |
| 97 | assert result["tips_built"] == 2 |
| 98 | assert result["tips_skipped"] == 0 |
| 99 | |
| 100 | |
| 101 | @pytest.mark.tier2 |
| 102 | async def test_fmc_07b_skips_tips_with_fresh_cache(db_session: AsyncSession) -> None: |
| 103 | """FMC_07b: prebuild skips ALL tips only when every tip shares the same cached mpack_id. |
| 104 | |
| 105 | When one tip is already cached with a DIFFERENT mpack_id than would be built for a |
| 106 | new tip, the prebuild must rebuild ALL tips together so that every tip ends up |
| 107 | pointing to the same combined mpack_id β required for the clone cache-hit check |
| 108 | (len(mpack_ids)==1 across all want tips). |
| 109 | """ |
| 110 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 111 | tip_cached = _fake_commit_id("tip-cached") |
| 112 | tip_new = _fake_commit_id("tip-new") |
| 113 | existing_mpack = _fake_mpack_id("existing") |
| 114 | new_mpack = _fake_mpack_id("new") |
| 115 | |
| 116 | # Pre-populate a fresh cache entry for tip_cached (different mpack_id from what |
| 117 | # the combined build will produce β diverged state). |
| 118 | cache_id = hashlib.sha256((repo.repo_id + tip_cached).encode()).hexdigest() |
| 119 | db_session.add(MusehubFetchMPackCache( |
| 120 | cache_id=cache_id, |
| 121 | repo_id=repo.repo_id, |
| 122 | tip_commit_id=tip_cached, |
| 123 | mpack_id=existing_mpack, |
| 124 | created_at=_now(), |
| 125 | expires_at=_now() + timedelta(days=7), |
| 126 | )) |
| 127 | # The handler reads live branch tips from MusehubBranch (Phase 1 fix). |
| 128 | await create_branch(db_session, repo.repo_id, name="main", head_commit_id=tip_cached) |
| 129 | await create_branch(db_session, repo.repo_id, name="dev", head_commit_id=tip_new) |
| 130 | job_id = await _insert_job(db_session, repo.repo_id, [tip_cached, tip_new]) |
| 131 | await db_session.commit() |
| 132 | |
| 133 | with patch( |
| 134 | "musehub.services.musehub_wire_fetch.wire_fetch_mpack", |
| 135 | new_callable=AsyncMock, |
| 136 | return_value={"mpack_id": new_mpack, "mpack_url": "https://r2.example/new", "commit_count": 2, "blob_count": 2}, |
| 137 | ) as mock_build: |
| 138 | result = await process_fetch_mpack_prebuild_job(db_session, job_id) |
| 139 | await db_session.commit() |
| 140 | |
| 141 | # Both tips must be rebuilt together so they share one mpack_id. |
| 142 | assert mock_build.call_count == 1 |
| 143 | actual_want = mock_build.call_args[1].get("want") or mock_build.call_args[0][2] |
| 144 | assert set(actual_want) == {tip_cached, tip_new}, ( |
| 145 | f"want must include ALL tips; got {actual_want}" |
| 146 | ) |
| 147 | assert result["tips_built"] == 2 |
| 148 | assert result["tips_skipped"] == 0 |
| 149 | |
| 150 | # The previously-cached entry must be updated to the new combined mpack_id. |
| 151 | cached_row = (await db_session.execute( |
| 152 | select(MusehubFetchMPackCache) |
| 153 | .where(MusehubFetchMPackCache.repo_id == repo.repo_id) |
| 154 | .where(MusehubFetchMPackCache.tip_commit_id == tip_cached) |
| 155 | )).scalar_one() |
| 156 | assert cached_row.mpack_id == new_mpack, ( |
| 157 | "existing cache entry must be updated to the combined mpack_id" |
| 158 | ) |
| 159 | |
| 160 | |
| 161 | @pytest.mark.tier2 |
| 162 | async def test_fmc_07c_empty_payload_is_a_noop(db_session: AsyncSession) -> None: |
| 163 | """FMC_07c: job with no tip_commit_ids returns zeros without calling wire_fetch_mpack.""" |
| 164 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 165 | job_id = await _insert_job(db_session, repo.repo_id, []) |
| 166 | await db_session.commit() |
| 167 | |
| 168 | with patch( |
| 169 | "musehub.services.musehub_wire_fetch.wire_fetch_mpack", |
| 170 | new_callable=AsyncMock, |
| 171 | ) as mock_build: |
| 172 | result = await process_fetch_mpack_prebuild_job(db_session, job_id) |
| 173 | |
| 174 | assert mock_build.call_count == 0 |
| 175 | assert result["tips_requested"] == 0 |
| 176 | assert result["tips_built"] == 0 |
| 177 | |
| 178 | |
| 179 | # ββ FMC_08 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 180 | |
| 181 | @pytest.mark.tier2 |
| 182 | async def test_fmc_08_cache_row_has_correct_fields(db_session: AsyncSession) -> None: |
| 183 | """FMC_08: running the handler end to end (through a REAL wire_fetch_mpack) writes a |
| 184 | cache row with matching repo_id, tip, and mpack_id. Cache rows are written inside |
| 185 | wire_fetch_mpack, so this exercises the full handler -> build -> cache-write chain |
| 186 | rather than mocking the build away. Only the expensive externals (DAG walk, mpack |
| 187 | bytes, storage) are stubbed β same pattern as FMC_14.""" |
| 188 | from types import SimpleNamespace |
| 189 | |
| 190 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 191 | tip = _fake_commit_id("integration-tip") |
| 192 | built_mpack = _fake_mpack_id("integration") |
| 193 | |
| 194 | # The handler reads live branch tips from MusehubBranch (Phase 1 fix). |
| 195 | await create_branch(db_session, repo.repo_id, name="main", head_commit_id=tip) |
| 196 | job_id = await _insert_job(db_session, repo.repo_id, [tip]) |
| 197 | await db_session.commit() |
| 198 | |
| 199 | mock_backend = AsyncMock() |
| 200 | mock_backend.put_mpack.return_value = None |
| 201 | mock_backend.presign_mpack_get.return_value = "https://r2.example/int" |
| 202 | mock_backend.delete.return_value = None |
| 203 | fake_proxy = SimpleNamespace(commit_id=tip, snapshot_id=None, parent_ids=[]) |
| 204 | |
| 205 | with patch("musehub.services.musehub_wire_fetch.get_backend", return_value=mock_backend), \ |
| 206 | patch("musehub.services.musehub_wire_fetch._walk_commit_delta", new_callable=AsyncMock, |
| 207 | return_value={tip: fake_proxy}), \ |
| 208 | patch("muse.core.mpack.build_wire_mpack", return_value=b"MUSE\x00fake-mpack"), \ |
| 209 | patch("musehub.services.musehub_wire_fetch.blob_id", return_value=built_mpack): |
| 210 | result = await process_fetch_mpack_prebuild_job(db_session, job_id) |
| 211 | await db_session.commit() |
| 212 | |
| 213 | assert result["tips_built"] == 1 |
| 214 | |
| 215 | row = (await db_session.execute( |
| 216 | select(MusehubFetchMPackCache) |
| 217 | .where(MusehubFetchMPackCache.repo_id == repo.repo_id) |
| 218 | .where(MusehubFetchMPackCache.tip_commit_id == tip) |
| 219 | )).scalar_one() |
| 220 | |
| 221 | assert row.repo_id == repo.repo_id |
| 222 | assert row.tip_commit_id == tip |
| 223 | assert row.mpack_id == built_mpack |
| 224 | assert row.expires_at > _now() |
| 225 | |
| 226 | |
| 227 | # ββ FMC_13 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 228 | |
| 229 | @pytest.mark.tier2 |
| 230 | async def test_fmc_13_cache_hit_returns_presigned_url_without_blob_load(db_session: AsyncSession) -> None: |
| 231 | """FMC_13: cache hit returns the presigned URL immediately; blob-load path is never entered.""" |
| 232 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 233 | tip = _fake_commit_id("fmc13-tip") |
| 234 | cached_mpack = _fake_mpack_id("fmc13-tip") |
| 235 | expected_url = "https://r2.example/cached-fmc13" |
| 236 | |
| 237 | # Pre-populate a fresh cache entry. |
| 238 | cache_id = hashlib.sha256((repo.repo_id + tip).encode()).hexdigest() |
| 239 | db_session.add(MusehubFetchMPackCache( |
| 240 | cache_id=cache_id, |
| 241 | repo_id=repo.repo_id, |
| 242 | tip_commit_id=tip, |
| 243 | mpack_id=cached_mpack, |
| 244 | created_at=_now(), |
| 245 | expires_at=_now() + timedelta(days=7), |
| 246 | )) |
| 247 | await db_session.commit() |
| 248 | |
| 249 | mock_backend = AsyncMock() |
| 250 | mock_backend.presign_mpack_get.return_value = expected_url |
| 251 | |
| 252 | with patch("musehub.services.musehub_wire_fetch.get_backend", return_value=mock_backend), \ |
| 253 | patch("musehub.services.musehub_wire_fetch._walk_commit_delta", new_callable=AsyncMock) as mock_walk: |
| 254 | result = await wire_fetch_mpack(db_session, repo.repo_id, want=[tip], have=[]) |
| 255 | |
| 256 | # The expensive DAG walk must never have been called. |
| 257 | mock_walk.assert_not_called() |
| 258 | # The returned URL must match the presigned URL for the cached mpack. |
| 259 | assert result["mpack_url"] == expected_url |
| 260 | assert result["mpack_id"] == cached_mpack |
| 261 | |
| 262 | |
| 263 | # ββ FMC_14 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 264 | |
| 265 | @pytest.mark.tier2 |
| 266 | async def test_fmc_14_cache_miss_builds_and_writes_cache_row(db_session: AsyncSession) -> None: |
| 267 | """FMC_14: on a cache miss, wire_fetch_mpack builds the mpack and writes a cache row.""" |
| 268 | from types import SimpleNamespace |
| 269 | |
| 270 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 271 | tip = _fake_commit_id("fmc14-tip") |
| 272 | built_mpack = _fake_mpack_id("fmc14-tip") |
| 273 | built_url = "https://r2.example/built-fmc14" |
| 274 | |
| 275 | # No cache row exists β this is a cold miss. |
| 276 | mock_backend = AsyncMock() |
| 277 | mock_backend.put_mpack.return_value = None |
| 278 | mock_backend.presign_mpack_get.return_value = built_url |
| 279 | mock_backend.delete.return_value = None |
| 280 | |
| 281 | # _walk_commit_delta returns one proxy commit with no snapshot so that |
| 282 | # wire_fetch_mpack proceeds to build an empty-but-valid mpack without |
| 283 | # needing real commit/snapshot/object rows in the DB. |
| 284 | fake_proxy = SimpleNamespace(commit_id=tip, snapshot_id=None, parent_ids=[]) |
| 285 | |
| 286 | with patch("musehub.services.musehub_wire_fetch.get_backend", return_value=mock_backend), \ |
| 287 | patch("musehub.services.musehub_wire_fetch._walk_commit_delta", new_callable=AsyncMock, |
| 288 | return_value={tip: fake_proxy}), \ |
| 289 | patch("muse.core.mpack.build_wire_mpack", return_value=b"MUSE\x00fake-mpack"), \ |
| 290 | patch("musehub.services.musehub_wire_fetch.blob_id", return_value=built_mpack): |
| 291 | result = await wire_fetch_mpack(db_session, repo.repo_id, want=[tip], have=[], force_build=True) |
| 292 | await db_session.commit() |
| 293 | |
| 294 | assert result["mpack_url"] == built_url |
| 295 | assert result["mpack_id"] == built_mpack |
| 296 | |
| 297 | # The cache row must have been written. |
| 298 | row = (await db_session.execute( |
| 299 | select(MusehubFetchMPackCache) |
| 300 | .where(MusehubFetchMPackCache.repo_id == repo.repo_id) |
| 301 | .where(MusehubFetchMPackCache.tip_commit_id == tip) |
| 302 | )).scalar_one() |
| 303 | assert row.mpack_id == built_mpack |
| 304 | assert row.expires_at > _now() |
| 305 | |
| 306 | |
| 307 | # ββ FMC_18 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 308 | |
| 309 | @pytest.mark.tier2 |
| 310 | async def test_fmc_18_enqueue_push_intel_creates_prebuild_job_with_branch_tips( |
| 311 | db_session: AsyncSession, |
| 312 | ) -> None: |
| 313 | """FMC_18: enqueue_push_intel enqueues fetch.mpack.prebuild with all branch tip commit IDs.""" |
| 314 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 315 | |
| 316 | tip_a = _fake_commit_id("branch-main") |
| 317 | tip_b = _fake_commit_id("branch-dev") |
| 318 | await create_branch(db_session, repo.repo_id, name="main", head_commit_id=tip_a) |
| 319 | await create_branch(db_session, repo.repo_id, name="dev", head_commit_id=tip_b) |
| 320 | |
| 321 | await enqueue_push_intel( |
| 322 | db_session, |
| 323 | repo.repo_id, |
| 324 | head=tip_a, |
| 325 | domain_id=None, |
| 326 | branch="main", |
| 327 | ) |
| 328 | await db_session.commit() |
| 329 | |
| 330 | job_row = (await db_session.execute( |
| 331 | select(MusehubBackgroundJob) |
| 332 | .where(MusehubBackgroundJob.repo_id == repo.repo_id) |
| 333 | .where(MusehubBackgroundJob.job_type == "fetch.mpack.prebuild") |
| 334 | .where(MusehubBackgroundJob.status == "pending") |
| 335 | )).scalar_one() |
| 336 | |
| 337 | tip_ids = set(job_row.payload.get("tip_commit_ids", [])) |
| 338 | assert tip_a in tip_ids, f"main tip {tip_a[:20]} missing from payload" |
| 339 | assert tip_b in tip_ids, f"dev tip {tip_b[:20]} missing from payload" |
| 340 | |
| 341 | |
| 342 | # ββ FMC_20 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 343 | |
| 344 | @pytest.mark.tier2 |
| 345 | async def test_fmc_20_gc_deletes_expired_rows_and_r2_objects_leaves_fresh_untouched( |
| 346 | db_session: AsyncSession, |
| 347 | ) -> None: |
| 348 | """FMC_20: gc_fetch_mpack_cache deletes expired rows + R2 objects; fresh rows survive.""" |
| 349 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 350 | |
| 351 | tip_expired_a = _fake_commit_id("expired-a") |
| 352 | tip_expired_b = _fake_commit_id("expired-b") |
| 353 | tip_fresh = _fake_commit_id("fresh") |
| 354 | |
| 355 | mpack_expired_a = _fake_mpack_id("expired-a") |
| 356 | mpack_expired_b = _fake_mpack_id("expired-b") |
| 357 | mpack_fresh = _fake_mpack_id("fresh") |
| 358 | |
| 359 | past = _now() - timedelta(days=1) |
| 360 | future = _now() + timedelta(days=6) |
| 361 | |
| 362 | for tip, mpack, exp in [ |
| 363 | (tip_expired_a, mpack_expired_a, past), |
| 364 | (tip_expired_b, mpack_expired_b, past), |
| 365 | (tip_fresh, mpack_fresh, future), |
| 366 | ]: |
| 367 | cache_id = hashlib.sha256((repo.repo_id + tip).encode()).hexdigest() |
| 368 | db_session.add(MusehubFetchMPackCache( |
| 369 | cache_id=cache_id, |
| 370 | repo_id=repo.repo_id, |
| 371 | tip_commit_id=tip, |
| 372 | mpack_id=mpack, |
| 373 | created_at=_now(), |
| 374 | expires_at=exp, |
| 375 | )) |
| 376 | await db_session.commit() |
| 377 | |
| 378 | mock_backend = AsyncMock() |
| 379 | mock_backend.delete.return_value = None |
| 380 | |
| 381 | with patch("musehub.services.musehub_gc.get_backend", return_value=mock_backend): |
| 382 | n_deleted = await gc_fetch_mpack_cache(db_session, repo.repo_id) |
| 383 | await db_session.commit() |
| 384 | |
| 385 | assert n_deleted == 2 |
| 386 | |
| 387 | # R2 delete called exactly once for each expired mpack β in any order. |
| 388 | deleted_mpack_ids = {call.args[0] for call in mock_backend.delete.call_args_list} |
| 389 | assert mpack_expired_a in deleted_mpack_ids |
| 390 | assert mpack_expired_b in deleted_mpack_ids |
| 391 | assert mpack_fresh not in deleted_mpack_ids |
| 392 | |
| 393 | # Expired rows gone from DB. |
| 394 | remaining = (await db_session.execute( |
| 395 | select(MusehubFetchMPackCache) |
| 396 | .where(MusehubFetchMPackCache.repo_id == repo.repo_id) |
| 397 | )).scalars().all() |
| 398 | remaining_tips = {r.tip_commit_id for r in remaining} |
| 399 | assert tip_expired_a not in remaining_tips |
| 400 | assert tip_expired_b not in remaining_tips |
| 401 | |
| 402 | # Fresh row survives. |
| 403 | assert tip_fresh in remaining_tips |
| 404 | |
| 405 | |
| 406 | # ββ FMC_21 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 407 | # Regression for musehub issue #39 (muse clone fails / hangs): a repo that goes |
| 408 | # stale (cache TTL expires with no further pushes) previously had no self-heal |
| 409 | # path β wire_fetch_mpack would raise MPackNotReadyError() forever on every |
| 410 | # retry, since nothing ever enqueued a new fetch.mpack.prebuild job for the |
| 411 | # current tips. Reproduced live against gabriel/muse on 2026-09-09: cache rows |
| 412 | # for both branch heads existed but had expired 5 days earlier, and the client |
| 413 | # retried the full 120s budget every single attempt with zero chance of success. |
| 414 | |
| 415 | @pytest.mark.tier2 |
| 416 | async def test_fmc_21_cache_miss_enqueues_prebuild_job(db_session: AsyncSession) -> None: |
| 417 | """FMC_21: on a real (non-force_build) cache miss, a fetch.mpack.prebuild job is enqueued.""" |
| 418 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 419 | tip = _fake_commit_id("fmc21-tip") |
| 420 | await create_branch(db_session, repo.repo_id, name="main", head_commit_id=tip) |
| 421 | |
| 422 | with pytest.raises(MPackNotReadyError): |
| 423 | await wire_fetch_mpack(db_session, repo.repo_id, want=[tip], have=[]) |
| 424 | await db_session.commit() |
| 425 | |
| 426 | job_row = (await db_session.execute( |
| 427 | select(MusehubBackgroundJob) |
| 428 | .where(MusehubBackgroundJob.repo_id == repo.repo_id) |
| 429 | .where(MusehubBackgroundJob.job_type == "fetch.mpack.prebuild") |
| 430 | .where(MusehubBackgroundJob.status == "pending") |
| 431 | )).scalar_one() |
| 432 | assert tip in set(job_row.payload.get("tip_commit_ids", [])) |
| 433 | |
| 434 | |
| 435 | @pytest.mark.tier2 |
| 436 | async def test_fmc_21b_repeated_cache_miss_does_not_duplicate_pending_job(db_session: AsyncSession) -> None: |
| 437 | """FMC_21b: a second MISS (e.g. the client's own retry loop) does not enqueue a duplicate job.""" |
| 438 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 439 | tip = _fake_commit_id("fmc21b-tip") |
| 440 | await create_branch(db_session, repo.repo_id, name="main", head_commit_id=tip) |
| 441 | |
| 442 | for _ in range(3): |
| 443 | with pytest.raises(MPackNotReadyError): |
| 444 | await wire_fetch_mpack(db_session, repo.repo_id, want=[tip], have=[]) |
| 445 | await db_session.commit() |
| 446 | |
| 447 | rows = (await db_session.execute( |
| 448 | select(MusehubBackgroundJob) |
| 449 | .where(MusehubBackgroundJob.repo_id == repo.repo_id) |
| 450 | .where(MusehubBackgroundJob.job_type == "fetch.mpack.prebuild") |
| 451 | .where(MusehubBackgroundJob.status == "pending") |
| 452 | )).scalars().all() |
| 453 | assert len(rows) == 1 |
| 454 | |
| 455 | |
| 456 | @pytest.mark.tier2 |
| 457 | async def test_fmc_21c_expired_cache_self_heals_end_to_end(db_session: AsyncSession) -> None: |
| 458 | """FMC_21c: expired cache -> MISS enqueues job -> running the job -> next fetch is a HIT. |
| 459 | |
| 460 | This is the exact sequence that was broken for gabriel/muse: a dormant |
| 461 | repo's cache entry ages past its TTL and, before this fix, could never |
| 462 | recover without a fresh push. |
| 463 | """ |
| 464 | from types import SimpleNamespace |
| 465 | |
| 466 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 467 | tip = _fake_commit_id("fmc21c-tip") |
| 468 | stale_mpack = _fake_mpack_id("fmc21c-stale") |
| 469 | await create_branch(db_session, repo.repo_id, name="main", head_commit_id=tip) |
| 470 | |
| 471 | # Seed an *expired* cache row -- exactly what we found live on staging. |
| 472 | cache_id = hashlib.sha256((repo.repo_id + tip).encode()).hexdigest() |
| 473 | db_session.add(MusehubFetchMPackCache( |
| 474 | cache_id=cache_id, |
| 475 | repo_id=repo.repo_id, |
| 476 | tip_commit_id=tip, |
| 477 | mpack_id=stale_mpack, |
| 478 | created_at=_now() - timedelta(days=8), |
| 479 | expires_at=_now() - timedelta(days=1), |
| 480 | )) |
| 481 | await db_session.commit() |
| 482 | |
| 483 | # Step 1 -- MISS (expired row doesn't count as a hit) must self-heal. |
| 484 | with pytest.raises(MPackNotReadyError): |
| 485 | await wire_fetch_mpack(db_session, repo.repo_id, want=[tip], have=[]) |
| 486 | await db_session.commit() |
| 487 | |
| 488 | job_row = (await db_session.execute( |
| 489 | select(MusehubBackgroundJob) |
| 490 | .where(MusehubBackgroundJob.repo_id == repo.repo_id) |
| 491 | .where(MusehubBackgroundJob.job_type == "fetch.mpack.prebuild") |
| 492 | .where(MusehubBackgroundJob.status == "pending") |
| 493 | )).scalar_one() |
| 494 | |
| 495 | # Step 2 -- run the job the worker would have picked up. |
| 496 | fresh_mpack = _fake_mpack_id("fmc21c-fresh") |
| 497 | fake_proxy = SimpleNamespace(commit_id=tip, snapshot_id=None, parent_ids=[]) |
| 498 | mock_backend = AsyncMock() |
| 499 | mock_backend.put_mpack.return_value = None |
| 500 | mock_backend.presign_mpack_get.return_value = "https://r2.example/fresh-fmc21c" |
| 501 | mock_backend.delete.return_value = None |
| 502 | |
| 503 | with patch("musehub.services.musehub_wire_fetch.get_backend", return_value=mock_backend), \ |
| 504 | patch("musehub.services.musehub_wire_fetch._walk_commit_delta", new_callable=AsyncMock, |
| 505 | return_value={tip: fake_proxy}), \ |
| 506 | patch("muse.core.mpack.build_wire_mpack", return_value=b"MUSE\x00fake-mpack"), \ |
| 507 | patch("musehub.services.musehub_wire_fetch.blob_id", return_value=fresh_mpack): |
| 508 | await process_fetch_mpack_prebuild_job(db_session, job_row.job_id) |
| 509 | await db_session.commit() |
| 510 | |
| 511 | # Step 3 -- the next fetch is now a real cache HIT on the freshly built mpack. |
| 512 | with patch("musehub.services.musehub_wire_fetch.get_backend", return_value=mock_backend), \ |
| 513 | patch("musehub.services.musehub_wire_fetch._walk_commit_delta", new_callable=AsyncMock) as mock_walk: |
| 514 | result = await wire_fetch_mpack(db_session, repo.repo_id, want=[tip], have=[]) |
| 515 | |
| 516 | mock_walk.assert_not_called() |
| 517 | assert result["mpack_id"] == fresh_mpack |
| 518 | assert result["mpack_id"] != stale_mpack |