feat(scheduler): order due sources by last_checked_at — most overdue first

select_due_sources returned rows in undefined order (Postgres-determined,
typically PK). At tick rates that outpace download-queue throughput, a
freshly-rerun source could keep getting re-queued ahead of one that's
still waiting for its first attempt this cycle. Operator-flagged
2026-05-30:

> if there are 8 hours before a source is due again and 40 full time
> downloads can happen in that period that means that there's a chance
> the first one to fire gets back into the download queue before item 41
> has a chance to get downloaded.

Added `ORDER BY last_checked_at ASC NULLS FIRST, id` to the due-source
SELECT. Never-checked sources go first, then longest-since-checked, then
ties broken by id. Combined with Celery's FIFO `download` queue, the
oldest-overdue source in each tick now reaches a worker before any
fresher one.

Test pins the ordering: a NULL-last_checked source, a 4-hour-overdue
source, and a 2-min-overdue source come back in that exact order from
select_due_sources.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-05-30 12:08:43 -04:00
parent ff9e96e0e2
commit 77f7a23410
2 changed files with 47 additions and 0 deletions
@@ -116,6 +116,13 @@ async def select_due_sources(session: AsyncSession) -> list[Source]:
whose platform is currently in a rate-limit cooldown are excluded — the whose platform is currently in a rate-limit cooldown are excluded — the
cooldown is the preventive half of the burst-prevention pair (per-source cooldown is the preventive half of the burst-prevention pair (per-source
consecutive_failures backoff handles the offending source itself). consecutive_failures backoff handles the offending source itself).
Ordering: last_checked_at ASC NULLS FIRST, then id. Never-checked
sources go first, then the longest-since-checked, so the most overdue
sources hit Celery's FIFO download queue first. Anti-starvation: if
queue throughput ever falls below the tick rate, a freshly-rerun source
can't keep cutting in line ahead of one that hasn't been checked at all.
Operator-confirmed 2026-05-30.
""" """
rows = (await session.execute( rows = (await session.execute(
select(Source) select(Source)
@@ -123,6 +130,7 @@ async def select_due_sources(session: AsyncSession) -> list[Source]:
.join(Artist, Source.artist_id == Artist.id) .join(Artist, Source.artist_id == Artist.id)
.where(Source.enabled.is_(True)) .where(Source.enabled.is_(True))
.where(Artist.auto_check.is_(True)) .where(Artist.auto_check.is_(True))
.order_by(Source.last_checked_at.asc().nulls_first(), Source.id)
)).scalars().all() )).scalars().all()
cooldowns = await _platforms_in_cooldown(session) cooldowns = await _platforms_in_cooldown(session)
+39
View File
@@ -233,6 +233,45 @@ async def test_select_ignores_expired_cooldown(db):
assert any(s.url == "https://cd-exp" for s in due) assert any(s.url == "https://cd-exp" for s in due)
@pytest.mark.asyncio
async def test_select_orders_most_overdue_first_then_id(db):
"""Within the due set, sources are ordered by last_checked_at ASC NULLS
FIRST. Combined with Celery FIFO on the download queue, the most-overdue
source in each tick reaches a worker first — preventing the 'a freshly-
rerun source keeps cutting in line ahead of one that's still waiting'
starvation pattern when queue throughput is below the tick population."""
artist = await _seed_artist(db, interval=60, name="order-test")
now = datetime.now(UTC)
# Newest check — least overdue (but still past its 60s interval).
s_new = Source(
artist_id=artist.id, platform="patreon", url="https://order-new",
enabled=True, consecutive_failures=0,
last_checked_at=now - timedelta(minutes=2),
)
# Oldest check — most overdue among checked sources.
s_old = Source(
artist_id=artist.id, platform="patreon", url="https://order-old",
enabled=True, consecutive_failures=0,
last_checked_at=now - timedelta(hours=4),
)
# Never checked — wins the ordering (NULLS FIRST).
s_never = Source(
artist_id=artist.id, platform="patreon", url="https://order-never",
enabled=True, consecutive_failures=0,
)
db.add_all([s_new, s_old, s_never])
await db.commit()
due = await select_due_sources(db)
# Filter to just our test seeds (concurrent tests may add others).
urls = [s.url for s in due if s.url.startswith("https://order-")]
assert urls == [
"https://order-never",
"https://order-old",
"https://order-new",
]
@pytest.mark.asyncio @pytest.mark.asyncio
async def test_set_platform_cooldown_upserts(db): async def test_set_platform_cooldown_upserts(db):
"""Calling set_platform_cooldown twice on the same platform updates """Calling set_platform_cooldown twice on the same platform updates