test_mwp8_cache_invalidation.py
python
sha256:d035733f21ccff27735fddebfbbe0ed24565a32a22db8de5885402262671ecd2
chore: bump version to 0.2.0rc15 for musehub#113 fix release
Sonnet 4.6
patch
15 days ago
| 1 | """MWP-8 — fetch-mpack cache invalidation (musehub#111). |
| 2 | |
| 3 | AUDIT VERDICT (Phase 0 — AU_01..AU_04): |
| 4 | ======================================== |
| 5 | All three repair endpoints (wire_repair_object, wire_repair_snapshot, |
| 6 | wire_repair_commit) and the force-push path (wire_push_unpack_mpack with |
| 7 | force=True) share the same gap: |
| 8 | |
| 9 | 1. repair-object (AU_01) — wire_repair_object performs upsert + session.commit() |
| 10 | but NEVER touches musehub_fetch_mpack_cache. A fresh cache row for the tip |
| 11 | that reaches the repaired object survives unchanged. A subsequent fresh-clone |
| 12 | request (want=[tip], have=[]) returns a HIT on the pre-repair mpack_id. |
| 13 | |
| 14 | 2. repair-snapshot (AU_02) — wire_repair_snapshot resets manifest_blob + |
| 15 | session.commit() but NEVER touches musehub_fetch_mpack_cache. Same HIT |
| 16 | on stale bytes. |
| 17 | |
| 18 | 3. repair-commit (AU_03) — wire_repair_commit force-overwrites identity fields + |
| 19 | session.commit() but NEVER touches musehub_fetch_mpack_cache. Same HIT. |
| 20 | This is the exact mechanism behind the gabriel/muse rc10 incident: the |
| 21 | repair fixed the DB row but the cache kept serving the corrupted mpack for |
| 22 | 7 more days. |
| 23 | |
| 24 | 4. force-push (AU_04) — wire_push_unpack_mpack(force=True) advances the |
| 25 | branch pointer and enqueues intel jobs but NEVER deletes the old-tip cache |
| 26 | row. A client that still knows the old tip gets a HIT on content that is |
| 27 | no longer on any branch. |
| 28 | |
| 29 | Expected post-fix behavior (implemented in Phases 2–3): |
| 30 | • AU_01..AU_03 → after repair, wire_fetch_mpack(want=[tip], have=[]) raises |
| 31 | MPackNotReadyError (cache row deleted in same transaction as repair). |
| 32 | • AU_04 → after force-push, (repo_id, old_tip) row is gone; the same |
| 33 | fetch raises MPackNotReadyError. |
| 34 | |
| 35 | Phase 2 (RP_04..RP_06) and Phase 3 (FP_02) replace these assertions with the |
| 36 | corrected contracts. |
| 37 | |
| 38 | Run: |
| 39 | python3 -m pytest tests/test_mwp8_cache_invalidation.py -q --tb=short |
| 40 | """ |
| 41 | from __future__ import annotations |
| 42 | |
| 43 | import hashlib |
| 44 | from datetime import datetime, timedelta, timezone |
| 45 | from typing import Any |
| 46 | from unittest.mock import AsyncMock, patch |
| 47 | |
| 48 | import pytest |
| 49 | from sqlalchemy import select |
| 50 | from sqlalchemy.ext.asyncio import AsyncSession |
| 51 | |
| 52 | from muse.core.ids import hash_commit, long_id |
| 53 | from muse.core.mpack import build_wire_mpack |
| 54 | from muse.core.types import blob_id, fake_id |
| 55 | from musehub.db.musehub_repo_models import ( |
| 56 | MusehubBranch, |
| 57 | MusehubCommit, |
| 58 | MusehubCommitRef, |
| 59 | MusehubFetchMPackCache, |
| 60 | MusehubObject, |
| 61 | MusehubObjectRef, |
| 62 | ) |
| 63 | from musehub.db.musehub_jobs_models import MusehubBackgroundJob |
| 64 | from musehub.muse_cli.snapshot import compute_snapshot_id |
| 65 | from musehub.services.musehub_wire_fetch import wire_fetch_mpack |
| 66 | from musehub.services.musehub_wire_push import ( |
| 67 | wire_push_unpack_mpack, |
| 68 | wire_repair_commit, |
| 69 | wire_repair_object, |
| 70 | wire_repair_snapshot, |
| 71 | ) |
| 72 | from musehub.services.musehub_wire_shared import MPackNotReadyError |
| 73 | from tests.factories import create_branch, create_repo |
| 74 | |
| 75 | |
| 76 | # ── helpers ─────────────────────────────────────────────────────────────────── |
| 77 | |
| 78 | |
| 79 | def _now() -> datetime: |
| 80 | return datetime.now(tz=timezone.utc) |
| 81 | |
| 82 | |
| 83 | async def _insert_fresh_cache_row( |
| 84 | session: AsyncSession, |
| 85 | repo_id: str, |
| 86 | tip_commit_id: str, |
| 87 | mpack_id: str, |
| 88 | seed: str, |
| 89 | ) -> MusehubFetchMPackCache: |
| 90 | """Seed a non-expired cache row (tip → mpack_id) into musehub_fetch_mpack_cache.""" |
| 91 | now = _now() |
| 92 | row = MusehubFetchMPackCache( |
| 93 | cache_id=fake_id(f"cache-{seed}"), |
| 94 | repo_id=repo_id, |
| 95 | tip_commit_id=tip_commit_id, |
| 96 | mpack_id=mpack_id, |
| 97 | created_at=now, |
| 98 | expires_at=now + timedelta(days=7), |
| 99 | ) |
| 100 | session.add(row) |
| 101 | await session.commit() |
| 102 | return row |
| 103 | |
| 104 | |
| 105 | async def _cache_rows_for_repo( |
| 106 | session: AsyncSession, |
| 107 | repo_id: str, |
| 108 | ) -> list[MusehubFetchMPackCache]: |
| 109 | """Return all cache rows for a given repo_id.""" |
| 110 | rows = await session.execute( |
| 111 | select(MusehubFetchMPackCache).where( |
| 112 | MusehubFetchMPackCache.repo_id == repo_id |
| 113 | ) |
| 114 | ) |
| 115 | return list(rows.scalars().all()) |
| 116 | |
| 117 | |
| 118 | def _stub_push_backend(monkeypatch: pytest.MonkeyPatch) -> AsyncMock: |
| 119 | """Patch the storage backend used by musehub_wire_push (repair calls backend.put).""" |
| 120 | backend = AsyncMock() |
| 121 | backend.put = AsyncMock(return_value="s3://test-bucket/object") |
| 122 | monkeypatch.setattr("musehub.services.musehub_wire_push.get_backend", lambda: backend) |
| 123 | return backend |
| 124 | |
| 125 | |
| 126 | def _stub_fetch_backend(monkeypatch: pytest.MonkeyPatch, presign_url: str = "https://r2.example/mpack") -> AsyncMock: |
| 127 | """Patch the storage backend used by musehub_wire_fetch (HIT calls backend.presign_mpack_get).""" |
| 128 | backend = AsyncMock() |
| 129 | backend.presign_mpack_get = AsyncMock(return_value=presign_url) |
| 130 | monkeypatch.setattr("musehub.services.musehub_wire_fetch.get_backend", lambda: backend) |
| 131 | return backend |
| 132 | |
| 133 | |
| 134 | def _stub_gc_backend(monkeypatch: pytest.MonkeyPatch) -> AsyncMock: |
| 135 | """Patch the storage backend used by musehub_gc (invalidate calls backend.delete).""" |
| 136 | backend = AsyncMock() |
| 137 | backend.delete = AsyncMock() |
| 138 | monkeypatch.setattr("musehub.services.musehub_gc.get_backend", lambda: backend) |
| 139 | return backend |
| 140 | |
| 141 | |
| 142 | def _build_force_push_backend( |
| 143 | monkeypatch: pytest.MonkeyPatch, |
| 144 | mpack_bytes: bytes, |
| 145 | ) -> AsyncMock: |
| 146 | """Patch the storage backend used inside wire_push_unpack_mpack. |
| 147 | |
| 148 | wire_push_unpack_mpack imports get_backend locally from musehub.storage.backends, |
| 149 | so we patch that site (not the module-level musehub.services.musehub_wire_push one). |
| 150 | The returned mock has get_mpack pre-configured to return mpack_bytes. |
| 151 | """ |
| 152 | backend = AsyncMock() |
| 153 | backend.get_mpack = AsyncMock(return_value=mpack_bytes) |
| 154 | backend.quarantine_mpack = AsyncMock() |
| 155 | backend.delete = AsyncMock() |
| 156 | monkeypatch.setattr("musehub.storage.backends.get_backend", lambda: backend) |
| 157 | return backend |
| 158 | |
| 159 | |
| 160 | async def _insert_object_row( |
| 161 | session: AsyncSession, |
| 162 | repo_id: str, |
| 163 | object_id: str, |
| 164 | ) -> None: |
| 165 | """Insert a minimal MusehubObject + ref so wire_repair_object can upsert it.""" |
| 166 | from sqlalchemy.dialects.postgresql import insert as _pg_insert |
| 167 | await session.execute( |
| 168 | _pg_insert(MusehubObject) |
| 169 | .values(object_id=object_id, path="test.bin", size_bytes=0, storage_uri="s3://bucket/x", content_cache=None) |
| 170 | .on_conflict_do_nothing(index_elements=["object_id"]) |
| 171 | ) |
| 172 | await session.execute( |
| 173 | _pg_insert(MusehubObjectRef) |
| 174 | .values(repo_id=repo_id, object_id=object_id) |
| 175 | .on_conflict_do_nothing(index_elements=["repo_id", "object_id"]) |
| 176 | ) |
| 177 | await session.commit() |
| 178 | |
| 179 | |
| 180 | def _build_valid_commit_dict( |
| 181 | *, |
| 182 | snapshot_id: str, |
| 183 | message: str, |
| 184 | committed_at: str, |
| 185 | author: str = "gabriel", |
| 186 | ) -> tuple[str, dict[str, Any]]: |
| 187 | """Compute a canonical (commit_id, wire_dict) pair that passes wire_repair_commit's hash check.""" |
| 188 | commit_id = hash_commit( |
| 189 | parent_ids=[], |
| 190 | snapshot_id=snapshot_id, |
| 191 | message=message, |
| 192 | committed_at_iso=committed_at, |
| 193 | author=author, |
| 194 | signer_public_key="", |
| 195 | ) |
| 196 | wire_dict = { |
| 197 | "commit_id": commit_id, |
| 198 | "branch": "main", |
| 199 | "snapshot_id": snapshot_id, |
| 200 | "message": message, |
| 201 | "committed_at": committed_at, |
| 202 | "parent_commit_id": None, |
| 203 | "parent2_commit_id": None, |
| 204 | "author": author, |
| 205 | "signer_public_key": "", |
| 206 | "signature": "", |
| 207 | "signer_key_id": "", |
| 208 | } |
| 209 | return commit_id, wire_dict |
| 210 | |
| 211 | |
| 212 | async def _insert_commit_row( |
| 213 | session: AsyncSession, |
| 214 | repo_id: str, |
| 215 | commit_id: str, |
| 216 | *, |
| 217 | snapshot_id: str, |
| 218 | message: str, |
| 219 | committed_at: str, |
| 220 | author: str = "gabriel", |
| 221 | ) -> MusehubCommit: |
| 222 | """Insert a MusehubCommit + MusehubCommitRef row that wire_repair_commit can look up.""" |
| 223 | ts = datetime.fromisoformat(committed_at.replace("Z", "+00:00")) |
| 224 | row = MusehubCommit( |
| 225 | commit_id=commit_id, |
| 226 | message=message, |
| 227 | author=author, |
| 228 | branch="main", |
| 229 | parent_ids=[], |
| 230 | snapshot_id=snapshot_id, |
| 231 | timestamp=ts, |
| 232 | signer_public_key="", |
| 233 | signature="", |
| 234 | signer_key_id="", |
| 235 | ) |
| 236 | session.add(row) |
| 237 | session.add(MusehubCommitRef(repo_id=repo_id, commit_id=commit_id)) |
| 238 | await session.commit() |
| 239 | return row |
| 240 | |
| 241 | |
| 242 | # ══════════════════════════════════════════════════════════════════════════════ |
| 243 | # RP_04 — repair-object busts cache (post-fix contract, converts AU_01) |
| 244 | # ══════════════════════════════════════════════════════════════════════════════ |
| 245 | |
| 246 | |
| 247 | @pytest.mark.asyncio |
| 248 | @pytest.mark.tier2 |
| 249 | async def test_RP_04_repair_object_busts_cache( |
| 250 | db_session: AsyncSession, |
| 251 | monkeypatch: pytest.MonkeyPatch, |
| 252 | ) -> None: |
| 253 | """RP_04 — wire_repair_object invalidates the cache in the same transaction. |
| 254 | |
| 255 | After repair, wire_fetch_mpack(want=[T], have=[]) raises MPackNotReadyError |
| 256 | (the cache row is gone — the next clone will MISS-rebuild from repaired rows). |
| 257 | Row count for repo_id is 0. |
| 258 | |
| 259 | Converts AU_01 (red-as-bug) to the post-fix correctness contract. |
| 260 | """ |
| 261 | content = b"rp04-test-object-mwp8" |
| 262 | object_id = blob_id(content) |
| 263 | tip_T = fake_id("tip-rp04") |
| 264 | mpack_M = fake_id("mpack-rp04") |
| 265 | |
| 266 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 267 | await _insert_object_row(db_session, repo.repo_id, object_id) |
| 268 | await _insert_fresh_cache_row(db_session, repo.repo_id, tip_T, mpack_M, seed="rp04") |
| 269 | |
| 270 | _stub_push_backend(monkeypatch) |
| 271 | _stub_gc_backend(monkeypatch) |
| 272 | |
| 273 | result = await wire_repair_object( |
| 274 | db_session, repo.repo_id, object_id, content, caller_id="gabriel" |
| 275 | ) |
| 276 | assert result["repaired"] is True |
| 277 | |
| 278 | # Post-fix: cache row is gone — fetch raises MPackNotReadyError (MISS) |
| 279 | with pytest.raises(MPackNotReadyError): |
| 280 | await wire_fetch_mpack( |
| 281 | db_session, repo.repo_id, want=[tip_T], have=[], force_build=False |
| 282 | ) |
| 283 | |
| 284 | remaining = await _cache_rows_for_repo(db_session, repo.repo_id) |
| 285 | assert len(remaining) == 0, "cache must be fully busted after repair-object" |
| 286 | |
| 287 | |
| 288 | # ══════════════════════════════════════════════════════════════════════════════ |
| 289 | # RP_05 — repair-snapshot busts cache (post-fix contract, converts AU_02) |
| 290 | # ══════════════════════════════════════════════════════════════════════════════ |
| 291 | |
| 292 | |
| 293 | @pytest.mark.asyncio |
| 294 | @pytest.mark.tier2 |
| 295 | async def test_RP_05_repair_snapshot_busts_cache( |
| 296 | db_session: AsyncSession, |
| 297 | monkeypatch: pytest.MonkeyPatch, |
| 298 | ) -> None: |
| 299 | """RP_05 — wire_repair_snapshot invalidates the cache in the same transaction. |
| 300 | |
| 301 | Converts AU_02 (red-as-bug) to the post-fix correctness contract. |
| 302 | """ |
| 303 | manifest: dict[str, str] = {"file.txt": fake_id("obj-rp05")} |
| 304 | directories: list[str] = [] |
| 305 | snapshot_id = compute_snapshot_id(manifest, directories) |
| 306 | tip_T = fake_id("tip-rp05") |
| 307 | mpack_M = fake_id("mpack-rp05") |
| 308 | |
| 309 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 310 | await _insert_fresh_cache_row(db_session, repo.repo_id, tip_T, mpack_M, seed="rp05") |
| 311 | |
| 312 | _stub_gc_backend(monkeypatch) |
| 313 | |
| 314 | result = await wire_repair_snapshot( |
| 315 | db_session, repo.repo_id, snapshot_id, manifest, directories, |
| 316 | caller_id="gabriel", |
| 317 | ) |
| 318 | assert result["repaired"] is True |
| 319 | |
| 320 | with pytest.raises(MPackNotReadyError): |
| 321 | await wire_fetch_mpack( |
| 322 | db_session, repo.repo_id, want=[tip_T], have=[], force_build=False |
| 323 | ) |
| 324 | |
| 325 | remaining = await _cache_rows_for_repo(db_session, repo.repo_id) |
| 326 | assert len(remaining) == 0, "cache must be fully busted after repair-snapshot" |
| 327 | |
| 328 | |
| 329 | # ══════════════════════════════════════════════════════════════════════════════ |
| 330 | # RP_06 — repair-commit busts cache (post-fix contract, converts AU_03) |
| 331 | # ══════════════════════════════════════════════════════════════════════════════ |
| 332 | |
| 333 | |
| 334 | @pytest.mark.asyncio |
| 335 | @pytest.mark.tier2 |
| 336 | async def test_RP_06_repair_commit_busts_cache( |
| 337 | db_session: AsyncSession, |
| 338 | monkeypatch: pytest.MonkeyPatch, |
| 339 | ) -> None: |
| 340 | """RP_06 — wire_repair_commit invalidates the cache in the same transaction. |
| 341 | |
| 342 | This is the direct fix for the gabriel/muse rc10 incident: repair-commit now |
| 343 | busts the cache so the next clone rebuilds from the corrected commit row. |
| 344 | |
| 345 | Converts AU_03 (red-as-bug) to the post-fix correctness contract. |
| 346 | """ |
| 347 | committed_at = "2026-01-01T00:00:00+00:00" |
| 348 | snapshot_id = fake_id("snap-rp06") |
| 349 | message = "rp06 post-fix repair-commit test" |
| 350 | commit_id, wire_dict = _build_valid_commit_dict( |
| 351 | snapshot_id=snapshot_id, |
| 352 | message=message, |
| 353 | committed_at=committed_at, |
| 354 | ) |
| 355 | |
| 356 | tip_T = commit_id |
| 357 | mpack_M = fake_id("mpack-rp06") |
| 358 | |
| 359 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 360 | await _insert_commit_row( |
| 361 | db_session, repo.repo_id, commit_id, |
| 362 | snapshot_id=snapshot_id, |
| 363 | message=message, |
| 364 | committed_at=committed_at, |
| 365 | ) |
| 366 | await _insert_fresh_cache_row(db_session, repo.repo_id, tip_T, mpack_M, seed="rp06") |
| 367 | |
| 368 | _stub_gc_backend(monkeypatch) |
| 369 | |
| 370 | result = await wire_repair_commit( |
| 371 | db_session, repo.repo_id, wire_dict, caller_id="gabriel" |
| 372 | ) |
| 373 | assert result["repaired"] is True |
| 374 | |
| 375 | with pytest.raises(MPackNotReadyError): |
| 376 | await wire_fetch_mpack( |
| 377 | db_session, repo.repo_id, want=[tip_T], have=[], force_build=False |
| 378 | ) |
| 379 | |
| 380 | remaining = await _cache_rows_for_repo(db_session, repo.repo_id) |
| 381 | assert len(remaining) == 0, "cache must be fully busted after repair-commit" |
| 382 | |
| 383 | |
| 384 | # ══════════════════════════════════════════════════════════════════════════════ |
| 385 | # RP_07 — atomicity guard: failed repair commit rolls back the invalidation |
| 386 | # ══════════════════════════════════════════════════════════════════════════════ |
| 387 | |
| 388 | |
| 389 | @pytest.mark.asyncio |
| 390 | @pytest.mark.tier2 |
| 391 | async def test_RP_07_repair_rollback_keeps_cache( |
| 392 | db_session: AsyncSession, |
| 393 | monkeypatch: pytest.MonkeyPatch, |
| 394 | ) -> None: |
| 395 | """RP_07 — if the repair's session.commit() fails, the invalidation rolls back too. |
| 396 | |
| 397 | No half-state: the cache row must survive a repair that fails to commit, |
| 398 | because both the repair write and the cache deletion are in the same transaction |
| 399 | (Design Decision D2). |
| 400 | |
| 401 | This prevents a scenario where: |
| 402 | - invalidate_fetch_mpack_cache deletes the row (no commit) |
| 403 | - session.commit() fails (e.g. constraint violation, DB error) |
| 404 | - the repair is rejected — but the cache is nuked → next clone is a MISS |
| 405 | on an un-repaired repo, hiding the failure from the operator. |
| 406 | |
| 407 | Uses wire_repair_object as the representative trigger. |
| 408 | """ |
| 409 | content = b"rp07-atomicity-guard-mwp8" |
| 410 | object_id = blob_id(content) |
| 411 | tip_T = fake_id("tip-rp07") |
| 412 | mpack_M = fake_id("mpack-rp07") |
| 413 | |
| 414 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 415 | repo_id = repo.repo_id # capture before any potential rollback |
| 416 | await _insert_object_row(db_session, repo_id, object_id) |
| 417 | await _insert_fresh_cache_row(db_session, repo_id, tip_T, mpack_M, seed="rp07") |
| 418 | |
| 419 | _stub_push_backend(monkeypatch) |
| 420 | _stub_gc_backend(monkeypatch) |
| 421 | |
| 422 | # Patch session.commit to raise after setup is complete — simulates a DB |
| 423 | # commit failure (e.g. a concurrent constraint violation). |
| 424 | original_commit = db_session.commit |
| 425 | |
| 426 | commit_calls = 0 |
| 427 | |
| 428 | async def failing_commit() -> None: |
| 429 | nonlocal commit_calls |
| 430 | commit_calls += 1 |
| 431 | raise RuntimeError("RP_07: simulated commit failure") |
| 432 | |
| 433 | monkeypatch.setattr(db_session, "commit", failing_commit) |
| 434 | |
| 435 | with pytest.raises(RuntimeError, match="RP_07"): |
| 436 | await wire_repair_object( |
| 437 | db_session, repo_id, object_id, content, caller_id="gabriel" |
| 438 | ) |
| 439 | |
| 440 | # Restore commit so we can rollback and then query |
| 441 | monkeypatch.setattr(db_session, "commit", original_commit) |
| 442 | await db_session.rollback() |
| 443 | |
| 444 | # The cache row must still exist — invalidation was not committed |
| 445 | remaining = await _cache_rows_for_repo(db_session, repo_id) |
| 446 | assert len(remaining) == 1, ( |
| 447 | "cache row must survive a repair whose commit failed — " |
| 448 | "invalidation must be in the same transaction (D2)" |
| 449 | ) |
| 450 | |
| 451 | |
| 452 | # ══════════════════════════════════════════════════════════════════════════════ |
| 453 | # FP_02 — force-push busts old + new tip rows (post-fix, converts AU_04) |
| 454 | # ══════════════════════════════════════════════════════════════════════════════ |
| 455 | |
| 456 | |
| 457 | @pytest.mark.asyncio |
| 458 | @pytest.mark.tier2 |
| 459 | async def test_FP_02_force_push_busts_old_and_new_tip( |
| 460 | db_session: AsyncSession, |
| 461 | monkeypatch: pytest.MonkeyPatch, |
| 462 | ) -> None: |
| 463 | """FP_02 — force-push deletes cache rows for both the old and new tip (FP_01 wiring). |
| 464 | |
| 465 | Scenario: |
| 466 | - Branch main currently points to T_old; a fresh cache row exists for T_old. |
| 467 | - A cache row for T_new also exists (e.g. T_new was once a prior branch head — |
| 468 | the hazard described in Background 'Trigger 2, hazard 2'). |
| 469 | - Force-push main → T_new. |
| 470 | |
| 471 | Post-fix contract: |
| 472 | - Both (repo_id, T_old) and (repo_id, T_new) rows are deleted in the same |
| 473 | transaction as the branch advance (Design Decision D1 tip-scoped, D2 same-tx). |
| 474 | - A subsequent clone of T_old raises MPackNotReadyError (MISS). |
| 475 | |
| 476 | Converts AU_04 (red-as-bug) to the post-fix correctness contract. |
| 477 | """ |
| 478 | from musehub.core.genesis import compute_branch_id |
| 479 | |
| 480 | mpack_bytes = build_wire_mpack({"objects": [], "commits": [], "snapshots": []}) |
| 481 | mpack_key = blob_id(mpack_bytes) |
| 482 | |
| 483 | T_old = fake_id("tip-old-fp02") |
| 484 | T_new = fake_id("tip-new-fp02") |
| 485 | mpack_old = fake_id("mpack-fp02-old") |
| 486 | mpack_new_cached = fake_id("mpack-fp02-new-cached") |
| 487 | |
| 488 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 489 | repo_id = repo.repo_id |
| 490 | |
| 491 | db_session.add(MusehubBranch( |
| 492 | branch_id=compute_branch_id(repo_id, "main"), |
| 493 | repo_id=repo_id, |
| 494 | name="main", |
| 495 | head_commit_id=T_old, |
| 496 | )) |
| 497 | await db_session.commit() |
| 498 | |
| 499 | # Seed cache rows for both tips (old + new) |
| 500 | await _insert_fresh_cache_row(db_session, repo_id, T_old, mpack_old, seed="fp02-old") |
| 501 | await _insert_fresh_cache_row(db_session, repo_id, T_new, mpack_new_cached, seed="fp02-new") |
| 502 | |
| 503 | _build_force_push_backend(monkeypatch, mpack_bytes) |
| 504 | _stub_gc_backend(monkeypatch) |
| 505 | |
| 506 | await wire_push_unpack_mpack( |
| 507 | db_session, |
| 508 | repo_id, |
| 509 | mpack_key=mpack_key, |
| 510 | pusher_id="gabriel", |
| 511 | branch="main", |
| 512 | head_commit_id=T_new, |
| 513 | force=True, |
| 514 | ) |
| 515 | |
| 516 | remaining = await _cache_rows_for_repo(db_session, repo_id) |
| 517 | remaining_tips = {r.tip_commit_id for r in remaining} |
| 518 | assert T_old not in remaining_tips, "(repo_id, T_old) must be gone after force-push" |
| 519 | assert T_new not in remaining_tips, "(repo_id, T_new) must be gone after force-push" |
| 520 | |
| 521 | # A clone of T_old is now a MISS |
| 522 | with pytest.raises(MPackNotReadyError): |
| 523 | await wire_fetch_mpack(db_session, repo_id, want=[T_old], have=[], force_build=False) |
| 524 | |
| 525 | |
| 526 | # ══════════════════════════════════════════════════════════════════════════════ |
| 527 | # FP_03 — normal (non-force) push does NOT invalidate sibling cache rows |
| 528 | # ══════════════════════════════════════════════════════════════════════════════ |
| 529 | |
| 530 | |
| 531 | @pytest.mark.asyncio |
| 532 | @pytest.mark.tier2 |
| 533 | async def test_FP_03_normal_push_does_not_invalidate( |
| 534 | db_session: AsyncSession, |
| 535 | monkeypatch: pytest.MonkeyPatch, |
| 536 | ) -> None: |
| 537 | """FP_03 — a non-force push must not invalidate sibling branch cache rows (Goal #3 / D3). |
| 538 | |
| 539 | Scenario: |
| 540 | - Repo has a 'dev' branch tip (T_dev) with a fresh cache row — the sibling. |
| 541 | - 'main' branch does not yet exist (first push to it). |
| 542 | - Push main → T_main (force=False, new branch creation). |
| 543 | |
| 544 | With current_head = None (new branch), the invalidation guard |
| 545 | `force and current_head and current_head != tip` is False on two counts |
| 546 | (force=False, current_head=None). The sibling T_dev row must survive intact. |
| 547 | |
| 548 | This asserts Design Decision D3: only force-push triggers invalidation; the |
| 549 | common fast-forward path pays zero invalidation cost. |
| 550 | """ |
| 551 | mpack_bytes = build_wire_mpack({"objects": [], "commits": [], "snapshots": []}) |
| 552 | mpack_key = blob_id(mpack_bytes) |
| 553 | |
| 554 | T_dev = fake_id("tip-dev-fp03") |
| 555 | T_main = fake_id("tip-main-fp03") |
| 556 | mpack_dev = fake_id("mpack-dev-fp03") |
| 557 | |
| 558 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 559 | repo_id = repo.repo_id |
| 560 | |
| 561 | # Seed the sibling dev branch cache row (no branch row needed — the cache |
| 562 | # row tests whether the sibling survives the push to main) |
| 563 | await _insert_fresh_cache_row(db_session, repo_id, T_dev, mpack_dev, seed="fp03-dev") |
| 564 | |
| 565 | _build_force_push_backend(monkeypatch, mpack_bytes) |
| 566 | _stub_gc_backend(monkeypatch) |
| 567 | |
| 568 | # Non-force push creating a new 'main' branch — current_head is None |
| 569 | await wire_push_unpack_mpack( |
| 570 | db_session, |
| 571 | repo_id, |
| 572 | mpack_key=mpack_key, |
| 573 | pusher_id="gabriel", |
| 574 | branch="main", |
| 575 | head_commit_id=T_main, |
| 576 | force=False, |
| 577 | ) |
| 578 | |
| 579 | remaining = await _cache_rows_for_repo(db_session, repo_id) |
| 580 | remaining_tips = {r.tip_commit_id for r in remaining} |
| 581 | assert T_dev in remaining_tips, "sibling dev cache row must survive a non-force push" |
| 582 | |
| 583 | |
| 584 | # ══════════════════════════════════════════════════════════════════════════════ |
| 585 | # FP_04 — force-push still enqueues a prebuild for T_new (loop closed) |
| 586 | # ══════════════════════════════════════════════════════════════════════════════ |
| 587 | |
| 588 | |
| 589 | @pytest.mark.asyncio |
| 590 | @pytest.mark.tier2 |
| 591 | async def test_FP_04_force_push_enqueues_prebuild_for_new_tip( |
| 592 | db_session: AsyncSession, |
| 593 | monkeypatch: pytest.MonkeyPatch, |
| 594 | ) -> None: |
| 595 | """FP_04 — after force-push, a fetch.mpack.prebuild job is enqueued for T_new. |
| 596 | |
| 597 | The invalidation (FP_01) deletes old cache rows; the existing enqueue_push_intel |
| 598 | path (unchanged by this ticket) then repopulates T_new with a fresh mpack. |
| 599 | |
| 600 | This test verifies the loop is still closed — force-push = delete stale rows + |
| 601 | enqueue rebuild — so the next clone is a fast HIT on correct content, not a slow |
| 602 | synchronous MISS-build or a stale HIT. |
| 603 | |
| 604 | Asserts Design Decision D4: correctness via deletion; speed via prebuild. |
| 605 | """ |
| 606 | from musehub.db.musehub_jobs_models import MusehubBackgroundJob |
| 607 | from musehub.core.genesis import compute_branch_id |
| 608 | |
| 609 | mpack_bytes = build_wire_mpack({"objects": [], "commits": [], "snapshots": []}) |
| 610 | mpack_key = blob_id(mpack_bytes) |
| 611 | |
| 612 | T_old = fake_id("tip-old-fp04") |
| 613 | T_new = fake_id("tip-new-fp04") |
| 614 | mpack_old = fake_id("mpack-fp04-old") |
| 615 | |
| 616 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 617 | repo_id = repo.repo_id |
| 618 | |
| 619 | db_session.add(MusehubBranch( |
| 620 | branch_id=compute_branch_id(repo_id, "main"), |
| 621 | repo_id=repo_id, |
| 622 | name="main", |
| 623 | head_commit_id=T_old, |
| 624 | )) |
| 625 | await db_session.commit() |
| 626 | |
| 627 | await _insert_fresh_cache_row(db_session, repo_id, T_old, mpack_old, seed="fp04-old") |
| 628 | |
| 629 | _build_force_push_backend(monkeypatch, mpack_bytes) |
| 630 | _stub_gc_backend(monkeypatch) |
| 631 | |
| 632 | await wire_push_unpack_mpack( |
| 633 | db_session, |
| 634 | repo_id, |
| 635 | mpack_key=mpack_key, |
| 636 | pusher_id="gabriel", |
| 637 | branch="main", |
| 638 | head_commit_id=T_new, |
| 639 | force=True, |
| 640 | ) |
| 641 | |
| 642 | # A fetch.mpack.prebuild job must exist for the repo (enqueued by enqueue_push_intel) |
| 643 | job_rows = (await db_session.execute( |
| 644 | select(MusehubBackgroundJob).where( |
| 645 | MusehubBackgroundJob.job_type == "fetch.mpack.prebuild", |
| 646 | MusehubBackgroundJob.repo_id == repo_id, |
| 647 | ) |
| 648 | )).scalars().all() |
| 649 | assert len(job_rows) >= 1, ( |
| 650 | "force-push must enqueue a fetch.mpack.prebuild job so T_new gets a fresh " |
| 651 | "cache row after invalidation (D4 — correctness via deletion, speed via prebuild)" |
| 652 | ) |
| 653 | |
| 654 | |
| 655 | # ══════════════════════════════════════════════════════════════════════════════ |
| 656 | # Phase 1 — The invalidation primitive (IV_01..IV_06) |
| 657 | # ══════════════════════════════════════════════════════════════════════════════ |
| 658 | |
| 659 | |
| 660 | # ── IV_01 ───────────────────────────────────────────────────────────────────── |
| 661 | |
| 662 | |
| 663 | @pytest.mark.asyncio |
| 664 | @pytest.mark.tier2 |
| 665 | async def test_IV_01_invalidate_all_deletes_every_row_for_repo( |
| 666 | db_session: AsyncSession, |
| 667 | monkeypatch: pytest.MonkeyPatch, |
| 668 | ) -> None: |
| 669 | """IV_01 — repo-wide invalidation deletes all rows for target repo; sibling repo untouched. |
| 670 | |
| 671 | Seed 3 rows for repo A (mix of fresh + expired) and 2 rows for repo B. |
| 672 | invalidate_fetch_mpack_cache(session, A) must return 3, leave A with 0 rows, |
| 673 | and leave B's 2 rows intact (cross-repo isolation). |
| 674 | """ |
| 675 | from musehub.services.musehub_gc import invalidate_fetch_mpack_cache |
| 676 | |
| 677 | _stub_push_backend(monkeypatch) |
| 678 | |
| 679 | repo_a = await create_repo(db_session, owner="gabriel", visibility="public") |
| 680 | repo_b = await create_repo(db_session, owner="gabriel", visibility="public") |
| 681 | |
| 682 | # 3 rows for A — mix of fresh and expired |
| 683 | for seed in ("iv01-a1", "iv01-a2", "iv01-a3"): |
| 684 | expired = seed.endswith("a3") |
| 685 | now = _now() |
| 686 | db_session.add(MusehubFetchMPackCache( |
| 687 | cache_id=fake_id(f"cache-{seed}"), |
| 688 | repo_id=repo_a.repo_id, |
| 689 | tip_commit_id=fake_id(f"tip-{seed}"), |
| 690 | mpack_id=fake_id(f"mpack-{seed}"), |
| 691 | created_at=now, |
| 692 | expires_at=now - timedelta(minutes=5) if expired else now + timedelta(days=7), |
| 693 | )) |
| 694 | await db_session.commit() |
| 695 | |
| 696 | # 2 rows for B — all fresh |
| 697 | for seed in ("iv01-b1", "iv01-b2"): |
| 698 | await _insert_fresh_cache_row(db_session, repo_b.repo_id, fake_id(f"tip-{seed}"), fake_id(f"mpack-{seed}"), seed=seed) |
| 699 | |
| 700 | deleted = await invalidate_fetch_mpack_cache(db_session, repo_a.repo_id) |
| 701 | await db_session.commit() |
| 702 | |
| 703 | assert deleted == 3, f"expected 3 deleted, got {deleted}" |
| 704 | assert len(await _cache_rows_for_repo(db_session, repo_a.repo_id)) == 0, "repo A must have 0 rows" |
| 705 | assert len(await _cache_rows_for_repo(db_session, repo_b.repo_id)) == 2, "repo B must be untouched" |
| 706 | |
| 707 | |
| 708 | # ── IV_02 ───────────────────────────────────────────────────────────────────── |
| 709 | |
| 710 | |
| 711 | @pytest.mark.asyncio |
| 712 | @pytest.mark.tier2 |
| 713 | async def test_IV_02_invalidate_deletes_regardless_of_expiry( |
| 714 | db_session: AsyncSession, |
| 715 | monkeypatch: pytest.MonkeyPatch, |
| 716 | ) -> None: |
| 717 | """IV_02 — invalidate deletes fresh (non-expired) rows; gc_fetch_mpack_cache does not. |
| 718 | |
| 719 | This is the key behavioral difference: gc only removes expired rows, invalidate |
| 720 | removes regardless of expires_at. |
| 721 | """ |
| 722 | from musehub.services.musehub_gc import gc_fetch_mpack_cache, invalidate_fetch_mpack_cache |
| 723 | |
| 724 | _stub_push_backend(monkeypatch) |
| 725 | |
| 726 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 727 | tip = fake_id("tip-iv02") |
| 728 | mpack_id = fake_id("mpack-iv02") |
| 729 | |
| 730 | # Insert a FRESH (non-expired) row |
| 731 | now = _now() |
| 732 | db_session.add(MusehubFetchMPackCache( |
| 733 | cache_id=fake_id("cache-iv02"), |
| 734 | repo_id=repo.repo_id, |
| 735 | tip_commit_id=tip, |
| 736 | mpack_id=mpack_id, |
| 737 | created_at=now, |
| 738 | expires_at=now + timedelta(days=7), |
| 739 | )) |
| 740 | await db_session.commit() |
| 741 | |
| 742 | # gc_fetch_mpack_cache must NOT delete the fresh row |
| 743 | gc_deleted = await gc_fetch_mpack_cache(db_session, repo.repo_id) |
| 744 | assert gc_deleted == 0, "gc must not delete fresh rows" |
| 745 | assert len(await _cache_rows_for_repo(db_session, repo.repo_id)) == 1, "row must survive gc" |
| 746 | |
| 747 | # invalidate_fetch_mpack_cache MUST delete the fresh row |
| 748 | inv_deleted = await invalidate_fetch_mpack_cache(db_session, repo.repo_id) |
| 749 | await db_session.commit() |
| 750 | |
| 751 | assert inv_deleted == 1, "invalidate must delete regardless of expiry" |
| 752 | assert len(await _cache_rows_for_repo(db_session, repo.repo_id)) == 0, "row must be gone after invalidate" |
| 753 | |
| 754 | |
| 755 | # ── IV_03 ───────────────────────────────────────────────────────────────────── |
| 756 | |
| 757 | |
| 758 | @pytest.mark.asyncio |
| 759 | @pytest.mark.tier2 |
| 760 | async def test_IV_03_invalidate_tip_scoped_leaves_siblings( |
| 761 | db_session: AsyncSession, |
| 762 | monkeypatch: pytest.MonkeyPatch, |
| 763 | ) -> None: |
| 764 | """IV_03 — tip-scoped invalidation deletes only the named tips; siblings survive. |
| 765 | |
| 766 | Seed rows for tips [X, Y, Z]. invalidate(..., tips=[X]) must delete only X; |
| 767 | Y and Z must be untouched. |
| 768 | """ |
| 769 | from musehub.services.musehub_gc import invalidate_fetch_mpack_cache |
| 770 | |
| 771 | _stub_push_backend(monkeypatch) |
| 772 | |
| 773 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 774 | tip_x = fake_id("tip-x-iv03") |
| 775 | tip_y = fake_id("tip-y-iv03") |
| 776 | tip_z = fake_id("tip-z-iv03") |
| 777 | |
| 778 | for tip, seed in ((tip_x, "iv03-x"), (tip_y, "iv03-y"), (tip_z, "iv03-z")): |
| 779 | await _insert_fresh_cache_row(db_session, repo.repo_id, tip, fake_id(f"mpack-{seed}"), seed=seed) |
| 780 | |
| 781 | deleted = await invalidate_fetch_mpack_cache(db_session, repo.repo_id, tips=[tip_x]) |
| 782 | await db_session.commit() |
| 783 | |
| 784 | assert deleted == 1, "only 1 row (X) must be deleted" |
| 785 | remaining = await _cache_rows_for_repo(db_session, repo.repo_id) |
| 786 | remaining_tips = {r.tip_commit_id for r in remaining} |
| 787 | assert tip_x not in remaining_tips, "X must be gone" |
| 788 | assert tip_y in remaining_tips, "Y must survive" |
| 789 | assert tip_z in remaining_tips, "Z must survive" |
| 790 | |
| 791 | |
| 792 | # ── IV_04 ───────────────────────────────────────────────────────────────────── |
| 793 | |
| 794 | |
| 795 | @pytest.mark.asyncio |
| 796 | @pytest.mark.tier2 |
| 797 | async def test_IV_04_invalidate_best_effort_r2_delete_swallows_failure( |
| 798 | db_session: AsyncSession, |
| 799 | monkeypatch: pytest.MonkeyPatch, |
| 800 | ) -> None: |
| 801 | """IV_04 — backend.delete raising must not propagate; DB row still deleted. |
| 802 | |
| 803 | Mirrors the best-effort pattern from test_bc2_gc_delete_failure_is_swallowed. |
| 804 | R2 unavailability must not block cache invalidation — the DB row is the source |
| 805 | of truth for the HIT gate, not the R2 object. |
| 806 | """ |
| 807 | from musehub.services.musehub_gc import invalidate_fetch_mpack_cache |
| 808 | |
| 809 | backend = AsyncMock() |
| 810 | backend.delete = AsyncMock(side_effect=RuntimeError("R2 unavailable")) |
| 811 | monkeypatch.setattr("musehub.services.musehub_gc.get_backend", lambda: backend) |
| 812 | |
| 813 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 814 | await _insert_fresh_cache_row(db_session, repo.repo_id, fake_id("tip-iv04"), fake_id("mpack-iv04"), seed="iv04") |
| 815 | |
| 816 | # Must not raise even though backend.delete raises |
| 817 | deleted = await invalidate_fetch_mpack_cache(db_session, repo.repo_id) |
| 818 | await db_session.commit() |
| 819 | |
| 820 | assert deleted == 1, "row must be deleted from DB even when R2 delete fails" |
| 821 | backend.delete.assert_awaited_once() |
| 822 | assert len(await _cache_rows_for_repo(db_session, repo.repo_id)) == 0, "DB row must be gone" |
| 823 | |
| 824 | |
| 825 | # ── IV_05 ───────────────────────────────────────────────────────────────────── |
| 826 | |
| 827 | |
| 828 | @pytest.mark.asyncio |
| 829 | @pytest.mark.tier2 |
| 830 | async def test_IV_05_invalidate_does_not_commit( |
| 831 | db_session: AsyncSession, |
| 832 | monkeypatch: pytest.MonkeyPatch, |
| 833 | ) -> None: |
| 834 | """IV_05 — invalidate_fetch_mpack_cache issues no COMMIT (Design Decision D2). |
| 835 | |
| 836 | The caller owns the transaction. Rows deleted within the open transaction |
| 837 | must be invisible after a session.rollback(), proving the deletion was not |
| 838 | committed by the primitive itself. |
| 839 | """ |
| 840 | from musehub.services.musehub_gc import invalidate_fetch_mpack_cache |
| 841 | |
| 842 | _stub_push_backend(monkeypatch) |
| 843 | |
| 844 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 845 | repo_id = repo.repo_id # capture before rollback; ORM objects expire on rollback |
| 846 | await _insert_fresh_cache_row(db_session, repo_id, fake_id("tip-iv05"), fake_id("mpack-iv05"), seed="iv05") |
| 847 | |
| 848 | # Rows are visible before invalidation |
| 849 | assert len(await _cache_rows_for_repo(db_session, repo_id)) == 1 |
| 850 | |
| 851 | deleted = await invalidate_fetch_mpack_cache(db_session, repo_id) |
| 852 | assert deleted == 1 |
| 853 | |
| 854 | # Within the open transaction, the row appears deleted |
| 855 | assert len(await _cache_rows_for_repo(db_session, repo_id)) == 0, "row deleted within transaction" |
| 856 | |
| 857 | # Roll back — if the primitive had committed, this would be a no-op and the row would stay gone. |
| 858 | # If the primitive did NOT commit (correct), the row is restored. |
| 859 | await db_session.rollback() |
| 860 | |
| 861 | assert len(await _cache_rows_for_repo(db_session, repo_id)) == 1, ( |
| 862 | "row must be restored after rollback — invalidate must not commit (D2)" |
| 863 | ) |
| 864 | |
| 865 | |
| 866 | # ── IV_06 ───────────────────────────────────────────────────────────────────── |
| 867 | |
| 868 | |
| 869 | @pytest.mark.asyncio |
| 870 | @pytest.mark.tier2 |
| 871 | async def test_IV_06_invalidate_empty_repo_is_noop_returns_zero( |
| 872 | db_session: AsyncSession, |
| 873 | monkeypatch: pytest.MonkeyPatch, |
| 874 | ) -> None: |
| 875 | """IV_06 — invalidating a repo with no cache rows is a no-op returning 0.""" |
| 876 | from musehub.services.musehub_gc import invalidate_fetch_mpack_cache |
| 877 | |
| 878 | _stub_push_backend(monkeypatch) |
| 879 | |
| 880 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 881 | |
| 882 | deleted = await invalidate_fetch_mpack_cache(db_session, repo.repo_id) |
| 883 | await db_session.commit() |
| 884 | |
| 885 | assert deleted == 0 |
| 886 | assert len(await _cache_rows_for_repo(db_session, repo.repo_id)) == 0 |
| 887 | |
| 888 | |
| 889 | # ══════════════════════════════════════════════════════════════════════════════ |
| 890 | # Phase 4 — Optional prebuild-on-repair + E2E clone-after-repair (EX_01..EX_03) |
| 891 | # ══════════════════════════════════════════════════════════════════════════════ |
| 892 | |
| 893 | |
| 894 | # ── EX_01 ───────────────────────────────────────────────────────────────────── |
| 895 | |
| 896 | |
| 897 | @pytest.mark.asyncio |
| 898 | @pytest.mark.tier2 |
| 899 | async def test_EX_01_repair_enqueues_prebuild( |
| 900 | db_session: AsyncSession, |
| 901 | monkeypatch: pytest.MonkeyPatch, |
| 902 | ) -> None: |
| 903 | """EX_01 — each repair function enqueues a fetch.mpack.prebuild job after invalidation. |
| 904 | |
| 905 | Closing the loop: after repair busts the cache, the worker rebuilds a fresh mpack |
| 906 | from the corrected rows so the NEXT clone is a fast HIT, not a slow synchronous |
| 907 | MISS-build (Design Decision D4 — correctness via deletion, speed via prebuild). |
| 908 | |
| 909 | Uses wire_repair_commit as the representative repair trigger. The branch must exist |
| 910 | so _enqueue_repair_prebuild has branch heads to cover. |
| 911 | """ |
| 912 | from musehub.core.genesis import compute_branch_id |
| 913 | |
| 914 | committed_at = "2026-01-01T00:00:00+00:00" |
| 915 | snapshot_id = fake_id("snap-ex01") |
| 916 | message = "ex01 prebuild-on-repair test" |
| 917 | commit_id, wire_dict = _build_valid_commit_dict( |
| 918 | snapshot_id=snapshot_id, |
| 919 | message=message, |
| 920 | committed_at=committed_at, |
| 921 | ) |
| 922 | |
| 923 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 924 | repo_id = repo.repo_id |
| 925 | |
| 926 | # Branch must exist so _enqueue_repair_prebuild finds a tip to cover |
| 927 | db_session.add(MusehubBranch( |
| 928 | branch_id=compute_branch_id(repo_id, "main"), |
| 929 | repo_id=repo_id, |
| 930 | name="main", |
| 931 | head_commit_id=commit_id, |
| 932 | )) |
| 933 | await _insert_commit_row( |
| 934 | db_session, repo_id, commit_id, |
| 935 | snapshot_id=snapshot_id, |
| 936 | message=message, |
| 937 | committed_at=committed_at, |
| 938 | ) |
| 939 | |
| 940 | _stub_gc_backend(monkeypatch) |
| 941 | |
| 942 | result = await wire_repair_commit(db_session, repo_id, wire_dict, caller_id="gabriel") |
| 943 | assert result["repaired"] is True |
| 944 | |
| 945 | # A fetch.mpack.prebuild job must be enqueued for the repo |
| 946 | job_rows = (await db_session.execute( |
| 947 | select(MusehubBackgroundJob).where( |
| 948 | MusehubBackgroundJob.job_type == "fetch.mpack.prebuild", |
| 949 | MusehubBackgroundJob.repo_id == repo_id, |
| 950 | ) |
| 951 | )).scalars().all() |
| 952 | assert len(job_rows) >= 1, ( |
| 953 | "repair must enqueue a fetch.mpack.prebuild job so the next clone " |
| 954 | "is a fast HIT on corrected content (EX_01 — D4)" |
| 955 | ) |
| 956 | |
| 957 | # The job payload must cover the branch tip |
| 958 | job = job_rows[0] |
| 959 | assert commit_id in job.payload.get("tip_commit_ids", []), ( |
| 960 | "prebuild job payload must include the branch tip so the worker " |
| 961 | "rebuilds the mpack for the corrected content" |
| 962 | ) |
| 963 | |
| 964 | |
| 965 | # ── EX_02 ───────────────────────────────────────────────────────────────────── |
| 966 | |
| 967 | |
| 968 | @pytest.mark.asyncio |
| 969 | @pytest.mark.tier2 |
| 970 | @pytest.mark.wire |
| 971 | async def test_EX_02_clone_after_repair_reflects_repair( |
| 972 | db_session: AsyncSession, |
| 973 | monkeypatch: pytest.MonkeyPatch, |
| 974 | ) -> None: |
| 975 | """EX_02 — load-bearing E2E: repair busts cache; second clone sees corrected rows. |
| 976 | |
| 977 | This is the regression test for the gabriel/muse rc10 incident: |
| 978 | - A push populates musehub_fetch_mpack_cache (simulated by seeding a fresh row). |
| 979 | - A first clone is a fast HIT returning mpack_M_stale. |
| 980 | - An operator calls repair-commit to fix a signer identity field. |
| 981 | - A second clone attempt raises MPackNotReadyError (MISS) — the stale mpack_M |
| 982 | is gone and the cache row was deleted in the same transaction as the repair. |
| 983 | - The corrected commit row is in the DB and will be served by the next rebuild. |
| 984 | - A fetch.mpack.prebuild job is enqueued so the rebuild happens before the |
| 985 | next real client request. |
| 986 | |
| 987 | Combines RP_06 (repair-commit invalidates) + EX_01 (prebuild enqueued) into a |
| 988 | single narrative to prove the full chain is wired correctly. |
| 989 | """ |
| 990 | from musehub.core.genesis import compute_branch_id |
| 991 | |
| 992 | committed_at = "2026-01-15T12:00:00+00:00" |
| 993 | snapshot_id = fake_id("snap-ex02") |
| 994 | message = "ex02 e2e repair test" |
| 995 | commit_id, wire_dict = _build_valid_commit_dict( |
| 996 | snapshot_id=snapshot_id, |
| 997 | message=message, |
| 998 | committed_at=committed_at, |
| 999 | ) |
| 1000 | |
| 1001 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 1002 | repo_id = repo.repo_id |
| 1003 | |
| 1004 | # Branch pointing to this commit (main tip = commit_id) |
| 1005 | db_session.add(MusehubBranch( |
| 1006 | branch_id=compute_branch_id(repo_id, "main"), |
| 1007 | repo_id=repo_id, |
| 1008 | name="main", |
| 1009 | head_commit_id=commit_id, |
| 1010 | )) |
| 1011 | await _insert_commit_row( |
| 1012 | db_session, repo_id, commit_id, |
| 1013 | snapshot_id=snapshot_id, |
| 1014 | message=message, |
| 1015 | committed_at=committed_at, |
| 1016 | ) |
| 1017 | |
| 1018 | # Seed the pre-repair mpack cache row (simulates state after a push + prebuild) |
| 1019 | mpack_stale = fake_id("mpack-stale-ex02") |
| 1020 | await _insert_fresh_cache_row(db_session, repo_id, commit_id, mpack_stale, seed="ex02") |
| 1021 | |
| 1022 | _stub_gc_backend(monkeypatch) |
| 1023 | _stub_fetch_backend(monkeypatch, presign_url="https://r2.example/stale-mpack") |
| 1024 | |
| 1025 | # ── First clone: cache HIT on pre-repair mpack ──────────────────────────── |
| 1026 | first_clone = await wire_fetch_mpack(db_session, repo_id, want=[commit_id], have=[]) |
| 1027 | assert first_clone["mpack_id"] == mpack_stale, "first clone must HIT the pre-repair mpack" |
| 1028 | |
| 1029 | # ── Repair: fix the commit's signer field ───────────────────────────────── |
| 1030 | result = await wire_repair_commit(db_session, repo_id, wire_dict, caller_id="gabriel") |
| 1031 | assert result["repaired"] is True |
| 1032 | |
| 1033 | # ── Second clone: MISS — cache was busted by the repair in the same tx ──── |
| 1034 | with pytest.raises(MPackNotReadyError): |
| 1035 | await wire_fetch_mpack(db_session, repo_id, want=[commit_id], have=[], force_build=False) |
| 1036 | |
| 1037 | # The stale cache row is gone |
| 1038 | remaining = await _cache_rows_for_repo(db_session, repo_id) |
| 1039 | assert len(remaining) == 0, "cache must be empty after repair" |
| 1040 | |
| 1041 | # The corrected commit row is in the DB — next rebuild will serve correct bytes |
| 1042 | from sqlalchemy import select as _select |
| 1043 | from musehub.db.musehub_repo_models import MusehubCommit as _MC |
| 1044 | commit_row = (await db_session.execute( |
| 1045 | _select(_MC).where(_MC.commit_id == commit_id) |
| 1046 | )).scalar_one() |
| 1047 | assert commit_row.signer_public_key == wire_dict["signer_public_key"], ( |
| 1048 | "the commit row must have the repaired signer_public_key — " |
| 1049 | "the next rebuild will serve this corrected content" |
| 1050 | ) |
| 1051 | |
| 1052 | # A prebuild job was enqueued to close the loop |
| 1053 | job_rows = (await db_session.execute( |
| 1054 | select(MusehubBackgroundJob).where( |
| 1055 | MusehubBackgroundJob.job_type == "fetch.mpack.prebuild", |
| 1056 | MusehubBackgroundJob.repo_id == repo_id, |
| 1057 | ) |
| 1058 | )).scalars().all() |
| 1059 | assert len(job_rows) >= 1, "repair must enqueue a prebuild so the next clone is a fast HIT" |
| 1060 | |
| 1061 | |
| 1062 | # ── EX_03 ───────────────────────────────────────────────────────────────────── |
| 1063 | |
| 1064 | |
| 1065 | @pytest.mark.asyncio |
| 1066 | @pytest.mark.tier2 |
| 1067 | async def test_EX_03_invalidation_is_logged( |
| 1068 | db_session: AsyncSession, |
| 1069 | monkeypatch: pytest.MonkeyPatch, |
| 1070 | caplog: pytest.LogCaptureFixture, |
| 1071 | ) -> None: |
| 1072 | """EX_03 — invalidation emits a log line naming repo, scope, and rows deleted. |
| 1073 | |
| 1074 | Operators need visibility into when the cache was busted — the gabriel/muse rc10 |
| 1075 | incident was invisible because there was no invalidation AND no log. This test |
| 1076 | asserts that after a repair trigger, a log record exists that an operator can |
| 1077 | grep for to confirm the bust happened. |
| 1078 | |
| 1079 | Checks the logger.info in invalidate_fetch_mpack_cache: |
| 1080 | "invalidate_fetch_mpack_cache: repo=... tips=all deleted=N rows" |
| 1081 | """ |
| 1082 | import logging |
| 1083 | |
| 1084 | committed_at = "2026-02-01T00:00:00+00:00" |
| 1085 | snapshot_id = fake_id("snap-ex03") |
| 1086 | message = "ex03 logging test" |
| 1087 | commit_id, wire_dict = _build_valid_commit_dict( |
| 1088 | snapshot_id=snapshot_id, |
| 1089 | message=message, |
| 1090 | committed_at=committed_at, |
| 1091 | ) |
| 1092 | |
| 1093 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 1094 | repo_id = repo.repo_id |
| 1095 | |
| 1096 | await _insert_commit_row( |
| 1097 | db_session, repo_id, commit_id, |
| 1098 | snapshot_id=snapshot_id, |
| 1099 | message=message, |
| 1100 | committed_at=committed_at, |
| 1101 | ) |
| 1102 | |
| 1103 | tip_T = commit_id |
| 1104 | await _insert_fresh_cache_row(db_session, repo_id, tip_T, fake_id("mpack-ex03"), seed="ex03") |
| 1105 | |
| 1106 | _stub_gc_backend(monkeypatch) |
| 1107 | |
| 1108 | with caplog.at_level(logging.INFO, logger="musehub.services.musehub_gc"): |
| 1109 | result = await wire_repair_commit(db_session, repo_id, wire_dict, caller_id="gabriel") |
| 1110 | |
| 1111 | assert result["repaired"] is True |
| 1112 | |
| 1113 | # The invalidation log line must appear, naming repo + deletion count |
| 1114 | gc_logs = [r.message for r in caplog.records if "invalidate_fetch_mpack_cache" in r.message] |
| 1115 | assert len(gc_logs) >= 1, ( |
| 1116 | "invalidate_fetch_mpack_cache must emit a log line so operators can confirm " |
| 1117 | "the bust happened (the rc10 incident was invisible — this log line is the fix)" |
| 1118 | ) |
| 1119 | log_line = gc_logs[0] |
| 1120 | assert "deleted=1" in log_line or "deleted" in log_line, ( |
| 1121 | f"log line must name the rows deleted; got: {log_line!r}" |
| 1122 | ) |
| 1123 | assert repo_id[:8] in log_line or "repo=" in log_line, ( |
| 1124 | f"log line must name the repo; got: {log_line!r}" |
| 1125 | ) |
File History
1 commit
sha256:d035733f21ccff27735fddebfbbe0ed24565a32a22db8de5885402262671ecd2
chore: bump version to 0.2.0rc15 for musehub#113 fix release
Sonnet 4.6
patch
15 days ago