fix(crawler): serialize sync_manga_chapters per-manga (0.35.6)

Two concurrent calls of sync_manga_chapters for the same manga both
read seen_keys, both run the drop UPDATE filtered on `NOT (key = ANY
$3)`, and the later commit can soft-drop a chapter the earlier had
just inserted (lost-update under MVCC). Today the cron tick is the
only caller and the daemon-level advisory lock keeps it single-flight,
but that lock is held on one pool connection and doesn't actually
serialize the *function*: any future caller (bookmark hook,
admin-triggered re-sync, parallel worker) would race against the cron.

Add `pg_advisory_xact_lock(hashtextextended(manga_id::text, 0))` at
the start of the transaction. Auto-releases on commit/rollback so a
panic mid-call can't strand the lock. Lock keyed per-manga so calls
for different mangas still parallelize.

Test sync_chapters_serializes_concurrent_calls_for_same_manga spawns
two tokio tasks calling the function concurrently with overlapping
chapter lists and asserts every chapter survives.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
MechaCat02
2026-05-29 20:45:01 +02:00
parent c6bb9160e3
commit 33f7e19077
5 changed files with 124 additions and 3 deletions

View File

@@ -328,6 +328,22 @@ pub async fn sync_manga_chapters(
chapters: &[SourceChapterRef],
) -> sqlx::Result<ChapterDiff> {
let mut tx = pool.begin().await?;
// Per-manga advisory lock. Two concurrent calls for the same manga
// would otherwise both read `seen_keys`, both run the drop UPDATE
// filtered on `NOT (key = ANY $3)`, and the later commit could soft-
// drop a chapter the earlier commit had just inserted (lost-update
// shape under MVCC). `pg_advisory_xact_lock` is scoped to this
// transaction: it auto-releases on COMMIT/ROLLBACK so a Rust-side
// panic mid-call doesn't strand the lock. The single-arg int8 form
// keyed by `hashtextextended(manga_id::text, 0)` shares Postgres'
// global advisory-lock namespace with `CRON_LOCK_KEY`, but collision
// is 2^-64 per pair (a UUID-derived hash hitting the fixed cron key
// is effectively impossible).
sqlx::query("SELECT pg_advisory_xact_lock(hashtextextended($1::text, 0))")
.bind(manga_id)
.execute(&mut *tx)
.await?;
let mut diff = ChapterDiff::default();
let seen_keys: Vec<String> = chapters
.iter()