feat(drive): stop writing in {file,folder}.user_id

plan:

    ```
    What step 5 is actually doing

    Goal: stop writing to storage.files.user_id / storage.folders.user_id on new INSERTs. This is a prerequisite for step 6 (dropping the column entirely). The column has existed since the pre-D0 single-owner era; every read that used to key on it has already been migrated to drive-membership grants over the last N days.

    What has to change together for this to be safe:

    ┌──────────────┬─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┬───────────────────────────────────────────────────────────────────────────────────────────────────────┐
    │    Piece     │                                                          What changes                                                           │                                                  Why                                                  │
    ├──────────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┼───────────────────────────────────────────────────────────────────────────────────────────────────────┤
    │ Migration    │ Swap storage.files uniqueness indexes from (folder_id, name, user_id) to (drive_id, folder_id, name)                            │ Otherwise, two new rows with user_id = NULL would both be allowed (PG treats NULLs as distinct) —     │
    │ (a)          │                                                                                                                                 │ uniqueness silently breaks                                                                            │
    ├──────────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┼───────────────────────────────────────────────────────────────────────────────────────────────────────┤
    │ Migration    │ ALTER user_id DROP NOT NULL on both tables                                                                                      │ Otherwise, dropping the INSERT bind violates NOT NULL and every write 500s                            │
    │ (b)          │                                                                                                                                 │                                                                                                       │
    ├──────────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┼───────────────────────────────────────────────────────────────────────────────────────────────────────┤
    │ Migration    │ Drop dead user_id-leading indexes                                                                                               │ Cheap cleanup — nothing scans them anymore                                                            │
    │ (c)          │                                                                                                                                 │                                                                                                       │
    ├──────────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┼───────────────────────────────────────────────────────────────────────────────────────────────────────┤
    │ PL/pgSQL (d) │ Rewrite storage.copy_folder_tree without user_id in the INSERT column list                                                      │ Cross-drive copy runs entirely in SQL, needs the same treatment                                       │
    ├──────────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┼───────────────────────────────────────────────────────────────────────────────────────────────────────┤
    │ Rust (e)     │ ~12 INSERT column-list drops across folder_db_repository, file_blob_write_repository, drive_pg_repository, dedup_service,       │ Actual write path                                                                                     │
    │              │ folder_service, load-seed.rs                                                                                                    │                                                                                                       │
    ├──────────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┼───────────────────────────────────────────────────────────────────────────────────────────────────────┤
    │ Rust (f)     │ Simplify folder_db_repository::create_folder's parent lookup to only fetch drive_id (was fetching (user_id, drive_id))          │ It's fetching a value it no longer needs                                                              │
    └──────────────┴─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┴───────────────────────────────────────────────────────────────────────────────────────────────────────┘
    ```
This commit is contained in:
Edouard Vanbelle
2026-07-02 21:18:01 +02:00
parent 25dcab8e85
commit 206a780494
22 changed files with 761 additions and 281 deletions
+16 -10
View File
@@ -526,10 +526,12 @@ async fn build_subtree(
// Level 0 — the subtree's "root" sits inside `mount_under`, not at
// parent_id=NULL. drive_id is inherited from the mount point.
// Post-D7: `user_id` omitted; `created_by` / `updated_by` bind to
// the seed caller.
let root: (Uuid,) = sqlx::query_as(
"INSERT INTO storage.folders
(name, parent_id, user_id, drive_id, created_by, updated_by)
SELECT $1, parent.id, $2, parent.drive_id, $2, $2
(name, parent_id, drive_id, created_by, updated_by)
SELECT $1, parent.id, parent.drive_id, $2, $2
FROM storage.folders parent
WHERE parent.id = $3::uuid
RETURNING id",
@@ -568,13 +570,13 @@ async fn build_subtree(
);
// drive_id derives from the parent folder — same pattern as
// file_blob_write_repository's resolve_owner_and_drive helper.
// Every parent in `current_level` already has a drive_id set,
// so the JOIN is guaranteed to find one.
// file_blob_write_repository's resolve_parent_drive helper.
// Post-D7: `user_id` omitted; provenance via `created_by` /
// `updated_by`.
let rows: Vec<(Uuid,)> = sqlx::query_as(
"INSERT INTO storage.folders
(name, parent_id, user_id, drive_id, created_by, updated_by)
SELECT f.name, f.parent_id, $1, parent.drive_id, $1, $1
(name, parent_id, drive_id, created_by, updated_by)
SELECT f.name, f.parent_id, parent.drive_id, $1, $1
FROM UNNEST($2::uuid[], $3::text[]) AS f(parent_id, name)
JOIN storage.folders parent ON parent.id = f.parent_id
RETURNING id",
@@ -653,13 +655,17 @@ async fn insert_files(
// Post-D0: storage.files.drive_id is NOT NULL — derive it from the
// parent folder (same pattern as file_blob_write_repository's
// INSERTs and the resolve_owner_and_drive helper). The folder's
// INSERTs and the resolve_parent_drive helper). The folder's
// drive_id was set during the M2 backfill or by the lifecycle hook
// for users provisioned after D0.
//
// Post-D7: `user_id` omitted; `created_by` / `updated_by` bind to
// the seed caller so provenance is preserved.
sqlx::query(
"INSERT INTO storage.files
(name, folder_id, user_id, drive_id, blob_hash, size, mime_type)
SELECT f.name, f.folder_id, $1, fo.drive_id, $2, 0, 'text/plain'
(name, folder_id, drive_id, blob_hash, size, mime_type,
created_by, updated_by)
SELECT f.name, f.folder_id, fo.drive_id, $2, 0, 'text/plain', $1, $1
FROM UNNEST($3::uuid[], $4::text[]) AS f(folder_id, name)
JOIN storage.folders fo ON fo.id = f.folder_id",
)