gabriel / musehub public
fix patch task/mpack-index-backfill #2 / 5
AI Agent gabriel · 6 days ago · Jun 1, 2026 · Diff

fix: blobs only in S3/mpack — remove commit/snapshot individual S3 writes

Architecture correction: S3/MinIO is for blobs only. - Commits: DB canonical (musehub_commits), no individual S3 writes - Snapshots: DB canonical (musehub_snapshots + manifest_blob), no S3 writes - Blobs: mpack:// in MinIO, byte-range indexed via musehub_mpack_index

Removed from all commit write paths (proposals, repository, sync, wire_push): commit_to_bytes + backend.put() + storage_uri population

Removed from all snapshot write paths (snapshot, wire_push repair+bulk): snapshot_to_bytes + backend.put() + storage_uri population

Simplified _snap_row_to_wire_s3 and _commit_to_wire_s3 to serve from DB (S3 reads are slower than DB and the data is already there).

New tests in test_blob_only_object_store.py cover the correct architecture. Deleted phase1/4/5 tests that tested the now-incorrect S3 writes for commits/snapshots.

138 tests passing.

sha256:f3995ec2c05c9c34b0e4d6e96349a811d0117a1c51d78096d757998ccb3c0520 sha
sha256:6dedad4b55964206bb6db3279998193bf7f089c9096bc7c57b4deb5dd8f92f25 parent
+36 ~13 −98 symbols
sha256:8b5e52788fe758ddd20195aad01eb72da3dd7cef966cbe725835adea37837092 snapshot
+36
symbols added
~13
symbols modified
−98
symbols removed
0
dead code introduced
Semantic Changes 147 symbols
~ tests/test_blob_only_object_store.py .py 29 symbols added
+ _make_repo function async_function _make_repo L45–55 ← moved from tests/test_mpack_index_stale_cleanup.py
+ _mock_backend function function _mock_backend L58–65
+ _uid function function _uid L40–42
+ AsyncMock import import AsyncMock L30–30
+ AsyncSession import import AsyncSession L29–29 ← moved from tests/test_object_store_invariant_phase5.py
+ MagicMock import import MagicMock L30–30
+ MusehubBranch import import MusehubBranch L34–34
+ MusehubCommit import import MusehubCommit L34–34
+ MusehubCommitGraph import import MusehubCommitGraph L34–34
+ MusehubCommitRef import import MusehubCommitRef L34–34
+ MusehubObject import import MusehubObject L34–34
+ MusehubRepo import import MusehubRepo L34–34
+ MusehubSnapshot import import MusehubSnapshot L34–34
+ MusehubSnapshotRef import import MusehubSnapshotRef L34–34
+ annotations import import annotations L24–24 ← moved from tests/test_object_store_invariant_phase5.py
+ blob_id import import blob_id L32–32
+ call import import call L30–30
+ compute_identity_id import import compute_identity_id L33–33
+ compute_repo_id import import compute_repo_id L33–33 ← moved from tests/test_object_store_invariant_phase5.py
+ datetime import import datetime L26–26 ← moved from tests/test_object_store_invariant_phase5.py
+ fake_id import import fake_id L32–32 ← moved from tests/test_object_store_invariant_phase1.py
+ patch import import patch L30–30
+ pytest import import pytest L27–27 ← moved from tests/test_object_store_invariant_phase5.py
+ select import import select L28–28
+ test_BOS1_commit_storage_uri_is_null_after_push function async_function test_BOS1_commit_storage_uri_is_null_after_push L73–100
+ test_BOS2_snapshot_storage_uri_is_null_after_push function async_function test_BOS2_snapshot_storage_uri_is_null_after_push L108–131
+ test_BOS3_blob_storage_uri_is_mpack_after_push function async_function test_BOS3_blob_storage_uri_is_mpack_after_push L139–168
+ test_BOS4_commits_served_from_db_not_s3 function async_function test_BOS4_commits_served_from_db_not_s3 L176–217
+ test_BOS6_merge_proposal_commit_storage_uri_is_null function async_function test_BOS6_merge_proposal_commit_storage_uri_is_null L225–269
~ tests/test_object_store_invariant_phase1.py .py 38 symbols removed
− TestCommitFilesToRepoObjectStore class class TestCommitFilesToRepoObjectStore L317–407
− test_commit_is_muse_binary_format method async_method test_commit_is_muse_binary_format L363–383
− test_commit_written_to_object_store method async_method test_commit_written_to_object_store L318–337
− test_snapshot_is_muse_binary_format method async_method test_snapshot_is_muse_binary_format L385–407
− test_snapshot_written_to_object_store method async_method test_snapshot_written_to_object_store L339–361
− TestMergeProposalObjectStore class class TestMergeProposalObjectStore L105–248
− test_merge_commit_object_is_muse_binary_format method async_method test_merge_commit_object_is_muse_binary_format L129–150
− test_merge_commit_storage_uri_populated method async_method test_merge_commit_storage_uri_populated L205–225
− test_merge_commit_written_to_object_store method async_method test_merge_commit_written_to_object_store L106–127
− test_merged_snapshot_is_muse_binary_format method async_method test_merged_snapshot_is_muse_binary_format L179–203
− test_merged_snapshot_storage_uri_populated method async_method test_merged_snapshot_storage_uri_populated L227–248
− test_merged_snapshot_written_to_object_store method async_method test_merged_snapshot_written_to_object_store L152–177
− TestRepoInitObjectStore class class TestRepoInitObjectStore L255–310
− test_init_commit_is_muse_binary_format method async_method test_init_commit_is_muse_binary_format L284–310
− test_init_commit_written_to_object_store method async_method test_init_commit_written_to_object_store L256–282
− TestUpsertSnapshotObjectStore class class TestUpsertSnapshotObjectStore L414–452
− test_snapshot_is_muse_binary_format method async_method test_snapshot_is_muse_binary_format L433–452
− test_snapshot_written_to_object_store method async_method test_snapshot_written_to_object_store L415–431
− _backend function function _backend L88–90
− _decode_header function function _decode_header L93–98
− _make_branch_with_commit function async_function _make_branch_with_commit L64–85
− _make_repo function async_function _make_repo L46–61
− _uid function function _uid L41–43
− AsyncSession import import AsyncSession L23–23
− MusehubBranch import import MusehubBranch L27–27
− MusehubCommit import import MusehubCommit L27–27
− MusehubCommitRef import import MusehubCommitRef L27–27
− MusehubRepo import import MusehubRepo L27–27
− MusehubSnapshot import import MusehubSnapshot L27–27
− MusehubSnapshotRef import import MusehubSnapshotRef L27–27
− annotations import import annotations L18–18
− blob_id import import blob_id L25–25
− compute_identity_id import import compute_identity_id L26–26
− compute_repo_id import import compute_repo_id L26–26
− datetime import import datetime L21–21
− fake_id import import fake_id L25–25 → moved to tests/test_blob_only_object_store.py
− json import import json L20–20
− pytest import import pytest L22–22
~ tests/test_object_store_invariant_phase5.py .py 34 symbols removed
− TestBackfillCommits class class TestBackfillCommits L91–149
− test_backfill_skips_commits_already_in_s3 method async_method test_backfill_skips_commits_already_in_s3 L132–149
− test_commit_s3_bytes_are_muse_binary_format method async_method test_commit_s3_bytes_are_muse_binary_format L110–130
− test_commit_storage_uri_populated_after_backfill method async_method test_commit_storage_uri_populated_after_backfill L92–108
− TestBackfillSnapshots class class TestBackfillSnapshots L156–278
− test_backfill_skips_snapshots_already_in_s3 method async_method test_backfill_skips_snapshots_already_in_s3 L260–278
− test_snapshot_delta_only_reconstructed_and_backfilled method async_method test_snapshot_delta_only_reconstructed_and_backfilled L217–258
− test_snapshot_manifest_correct_after_backfill method async_method test_snapshot_manifest_correct_after_backfill L198–215
− test_snapshot_s3_bytes_are_muse_binary_format method async_method test_snapshot_s3_bytes_are_muse_binary_format L175–196
− test_snapshot_storage_uri_populated_after_backfill method async_method test_snapshot_storage_uri_populated_after_backfill L157–173
− TestBackfillStats class class TestBackfillStats L285–315
− test_idempotent_second_run_writes_nothing method async_method test_idempotent_second_run_writes_nothing L295–315
− test_stats_zero_when_nothing_to_backfill method async_method test_stats_zero_when_nothing_to_backfill L286–293
− _backend function function _backend L39–41
− _make_legacy_commit function function _make_legacy_commit L58–70
− _make_legacy_snapshot function function _make_legacy_snapshot L73–84
− _make_repo function async_function _make_repo L44–55
− _uid function function _uid L35–36
− AsyncSession import import AsyncSession L21–21 → moved to tests/test_blob_only_object_store.py
− MusehubCommit import import MusehubCommit L26–26
− MusehubCommitRef import import MusehubCommitRef L26–26
− MusehubRepo import import MusehubRepo L26–26
− MusehubSnapshot import import MusehubSnapshot L26–26
− MusehubSnapshotRef import import MusehubSnapshotRef L26–26
− StrDict import import StrDict L25–25
− annotations import import annotations L14–14 → moved to tests/test_blob_only_object_store.py
− compute_identity_id import import compute_identity_id L24–24
− compute_repo_id import import compute_repo_id L24–24 → moved to tests/test_blob_only_object_store.py
− datetime import import datetime L17–17 → moved to tests/test_blob_only_object_store.py
− fake_id import import fake_id L23–23
− json import import json L16–16
− msgpack import import msgpack L19–19
− pytest import import pytest L20–20 → moved to tests/test_blob_only_object_store.py
− secrets import import secrets L18–18
~ deploy/enqueue_mpack_index_backfill.py .py 2 symbols modified
~ _get_unindexed_mpacks
~ main
~ musehub/services/musehub_proposals.py .py 1 symbol modified
~ merge_proposal
~ musehub/services/musehub_repository.py .py 1 symbol modified
~ create_repo
~ musehub/services/musehub_snapshot.py .py 1 symbol modified
~ upsert_snapshot_entries
~ musehub/services/musehub_sync.py .py 2 symbols modified
~ commit_files_to_repo
~ ingest_push
~ musehub/services/musehub_wire_push.py .py 1 symbol added, 2 symbols modified
+ purge_stale_mpack_index_entries function async_function purge_stale_mpack_index_entries L927–968
~ wire_push_unpack_mpack
~ wire_repair_snapshot
~ musehub/services/musehub_wire_shared.py .py 2 symbols modified
~ _commit_to_wire_s3
~ _snap_row_to_wire_s3
~ musehub/storage/backends.py .py 1 symbol added, 1 symbol modified
+ exists_mpack method async_method exists_mpack L427–440
~ BlobBackend
~ tests/test_mpack_index_always_enqueued.py .py 1 symbol added
+ test_MIE5_mpack_index_payload_contains_mpack_key function async_function test_MIE5_mpack_index_payload_contains_mpack_key L105–143
· tests/test_mpack_index_stale_cleanup.py .py
~ tests/test_mpack_index_stale_cleanup.py .py moved from tests/test_object_store_invariant_phase4.py; 4 added, 26 removed, 1 modified
− TestWireFetchCommitFromS3 class class TestWireFetchCommitFromS3 L236–294
− test_commit_s3_bytes_decode_correctly method async_method test_commit_s3_bytes_decode_correctly L251–275
− test_commit_storage_uri_is_set method async_method test_commit_storage_uri_is_set L237–249
− test_wire_fetch_returns_commit method async_method test_wire_fetch_returns_commit L277–294
− TestWireFetchFallback class class TestWireFetchFallback L301–349
− test_snapshot_fallback_when_no_storage_uri method async_method test_snapshot_fallback_when_no_storage_uri L302–349
− TestWireFetchSnapshotFromS3 class class TestWireFetchSnapshotFromS3 L126–229
− test_snapshot_manifest_served_from_s3 method async_method test_snapshot_manifest_served_from_s3 L127–158
− test_snapshot_storage_uri_is_set method async_method test_snapshot_storage_uri_is_set L217–229
− test_snapshot_wire_manifest_correct_even_when_manifest_blob_zeroed method async_method test_snapshot_wire_manifest_correct_even_when_manifest_blob_zeroed L160–215
− _backend function function _backend L43–45
− _make_commit_with_snapshot function async_function _make_commit_with_snapshot L61–119
− _make_repo function async_function _make_repo L48–58 → moved to tests/test_blob_only_object_store.py
− _uid function function _uid L39–40
− MusehubBranch import import MusehubBranch L27–27
− MusehubCommit import import MusehubCommit L27–27
− MusehubCommitGraph import import MusehubCommitGraph L27–27
− MusehubCommitRef import import MusehubCommitRef L27–27
− MusehubRepo import import MusehubRepo L27–27
− MusehubSnapshot import import MusehubSnapshot L27–27
− MusehubSnapshotRef import import MusehubSnapshotRef L27–27
− compute_branch_id import import compute_branch_id L26–26
− compute_identity_id import import compute_identity_id L26–26
− compute_repo_id import import compute_repo_id L26–26
− json import import json L19–19
− secrets import import secrets L21–21
+ select import import select L22–22
+ test_SC1_purge_removes_stale_entries function async_function test_SC1_purge_removes_stale_entries L33–63
+ test_SC2_purge_keeps_live_entries function async_function test_SC2_purge_keeps_live_entries L71–98
+ test_SC3_purge_returns_accurate_counts function async_function test_SC3_purge_returns_accurate_counts L106–135
~ MusehubMPackIndex
Older fix: skip mpacks gone from storage (GCed) in backfill script sha256:6dedad4b55964206bb6db3279998193bf7f089c9096bc7c57b4deb5dd8f92f25 All commits Newer chore: carry uncommitted changes (mpack_key in enqueue_push_inte… sha256:e652528bc7ab28fc0a6799df687779ee296be645a7e1b4120e271df96aebf20a

0 comments

No comments yet. Be the first to start the discussion.

To add a comment, use the Muse CLI: muse hub commit comment sha256:f3995ec2c05c9c34b0e4d6e96349a811d0117a1c51d78096d757998ccb3c0520 --body "your comment"