fix: index the chat queries that make large SQLite instances unusable (#27663)

The timer scheduler polls once a second and cancels on every message send and chat open, the sidebar lists chats ordered by `updated_at`, and the folder badges count unread chats per folder. None of those could be served by an index, so each call read most of the `chat` table, and because `meta` sits after the chat payload column SQLite had to walk every row's overflow pages to get there. On a large history that stalls the sidebar, every chat switch and every send, and the idle poll alone burns about a quarter of a CPU core.

Timers now keep their due time in a dedicated `chat.timer_at` column behind a partial index, and the chat list, unread and unfinished-reply queries each get an index matching their filter and ordering. Existing pending timers are backfilled from their meta by the migration. Dropping the `internal` and `type` checks also makes a forked timer chat inert, where a fork used to copy `meta` verbatim and become a second claim target that could fire a duplicate timer.

Measured on SQLite, same rows returned:

| query | before | after |
|---|---|---|
| idle timer poll (2000 chats, 0.43 GB) | 170 ms | 0.04 ms |
| cancel on send and chat open (4000 chats, 377 MB) | 200 ms | 0.04 ms |
| sidebar chat list (15000 chats, 1.26 GB) | 157 ms | 1.8 ms |
| folder unread badges (15000 chats, 1.4 GB) | 54 ms | 0.2 ms |

PostgreSQL 17 serves all of them as index-only scans with no sort node. Exercised through fresh install, upgrade with seeded data, downgrade and re-upgrade on SQLite and PostgreSQL 17.

Fixes #27622
This commit is contained in:
Classic298
2026-08-23 16:11:54 -05:00
committed by GitHub
parent 29e8d7db67
commit 16c2a9eda4
4 changed files with 103 additions and 9 deletions
@@ -0,0 +1,70 @@
"""add chat timer_at and chat list, unread and timer indexes
Revision ID: d4c1a8e37b62
Revises: 6d09d1bf1f23
Create Date: 2026-08-23 18:05:12.441907
"""
from collections.abc import Sequence
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision: str = 'd4c1a8e37b62'
down_revision: str | None = '6d09d1bf1f23'
branch_labels: str | Sequence[str] | None = None
depends_on: str | Sequence[str] | None = None
def upgrade() -> None:
op.add_column('chat', sa.Column('timer_at', sa.BigInteger(), nullable=True))
op.create_index(
'timer_at_idx',
'chat',
['timer_at'],
sqlite_where=sa.text('timer_at IS NOT NULL'),
postgresql_where=sa.text('timer_at IS NOT NULL'),
)
# Timers created before this migration carry their due time in meta only, and would never fire.
chat = sa.table(
'chat', sa.column('id', sa.String), sa.column('meta', sa.JSON), sa.column('timer_at', sa.BigInteger)
)
conn = op.get_bind()
pending = conn.execute(
sa.select(chat.c.id, chat.c.meta)
.where(chat.c.meta['type'].as_string() == 'timer')
.where(chat.c.meta['status'].as_string() == 'pending')
).all()
for chat_id, meta in pending:
try: # imported chats can carry any meta, and a non-numeric due time must not abort the migration
due_at = int(meta.get('timer_at'))
except (TypeError, ValueError):
continue
conn.execute(chat.update().where(chat.c.id == chat_id).values(timer_at=due_at))
op.create_index('user_id_updated_at_id_idx', 'chat', ['user_id', sa.text('updated_at DESC'), 'id'])
op.create_index(
'user_id_timer_at_idx',
'chat',
['user_id', 'timer_at'],
sqlite_where=sa.text('timer_at IS NOT NULL'),
postgresql_where=sa.text('timer_at IS NOT NULL'),
)
op.create_index(
'user_id_folder_unread_idx',
'chat',
['user_id', 'folder_id', 'archived', 'updated_at', 'last_read_at', 'id'],
)
op.create_index('chat_message_chat_role_done_idx', 'chat_message', ['chat_id', 'role', 'done'])
def downgrade() -> None:
op.drop_index('chat_message_chat_role_done_idx', table_name='chat_message')
op.drop_index('user_id_folder_unread_idx', table_name='chat')
op.drop_index('user_id_timer_at_idx', table_name='chat')
op.drop_index('user_id_updated_at_id_idx', table_name='chat')
op.drop_index('timer_at_idx', table_name='chat')
op.drop_column('chat', 'timer_at')
@@ -168,6 +168,7 @@ class ChatMessage(Base):
Index('chat_message_chat_parent_idx', 'chat_id', 'parent_id'),
Index('chat_message_model_created_idx', 'model_id', 'created_at'),
Index('chat_message_user_created_idx', 'user_id', 'created_at'),
Index('chat_message_chat_role_done_idx', 'chat_id', 'role', 'done'), # unfinished-assistant probe
)
+21
View File
@@ -149,6 +149,7 @@ class Chat(Base): # database table mapping for chat entity
current_message_id = Column(Text, nullable=True)
last_read_at = Column(BigInteger, nullable=True)
timer_at = Column(BigInteger, nullable=True) # ns due time, set only while a timer chat waits to be claimed
__table_args__ = (
# Performance indexes for common queries
@@ -157,6 +158,23 @@ class Chat(Base): # database table mapping for chat entity
Index('user_id_archived_idx', 'user_id', 'archived'),
Index('updated_at_user_id_idx', 'updated_at', 'user_id'),
Index('folder_id_user_id_idx', 'folder_id', 'user_id'),
Index('user_id_updated_at_id_idx', 'user_id', updated_at.desc(), 'id'),
Index(
'timer_at_idx',
'timer_at',
sqlite_where=text('timer_at IS NOT NULL'),
postgresql_where=text('timer_at IS NOT NULL'),
),
# timer_at key column turns the IS NOT NULL into a seek, so this beats the plain user_id indexes
Index(
'user_id_timer_at_idx',
'user_id',
'timer_at',
sqlite_where=text('timer_at IS NOT NULL'),
postgresql_where=text('timer_at IS NOT NULL'),
),
# covering index: lets SQLite serve count_unread_by_folder_ids without reading chat rows
Index('user_id_folder_unread_idx', 'user_id', 'folder_id', 'archived', 'updated_at', 'last_read_at', 'id'),
)
@@ -187,6 +205,7 @@ class ChatModel(BaseModel):
current_message_id: str | None = None
last_read_at: int | None = None
timer_at: int | None = None
@field_validator('variables', mode='before')
@classmethod
@@ -502,6 +521,7 @@ class ChatTable:
db: AsyncSession | None = None,
*,
internal_meta: dict | None = None,
timer_at: int | None = None,
) -> ChatModel | None:
async with get_async_db_context(db) as session:
chat = ChatModel(
@@ -514,6 +534,7 @@ class ChatTable:
'chat': self._clean_null_bytes(form_data.chat),
'folder_id': form_data.folder_id,
'meta': internal_meta or {},
'timer_at': timer_at,
'variables': form_data.variables or {},
'current_message_id': self.get_current_message_id(form_data.chat),
'created_at': int(time.time()),
+11 -9
View File
@@ -152,6 +152,7 @@ async def create_timer(
'cancel_on': selected_events,
'run': run,
},
timer_at=due_at,
)
if not chat:
return 'Error: failed to create timer.'
@@ -171,20 +172,18 @@ async def claim_due_timers(now_ns: int, limit: int = 10) -> list[tuple[str, str]
async with get_async_db() as db:
stmt = (
select(Chat)
.where(Chat.meta['internal'].as_boolean().is_(True))
.where(Chat.meta['type'].as_string() == 'timer')
.where(Chat.timer_at <= now_ns)
.where(Chat.meta['status'].as_string() == 'pending')
.order_by(Chat.timer_at)
.limit(limit)
)
if db.bind.dialect.name == 'postgresql':
stmt = stmt.with_for_update(skip_locked=True)
result = await db.execute(stmt)
rows = [row for row in result.scalars().all() if int((row.meta or {}).get('timer_at') or 0) <= now_ns]
rows.sort(key=lambda row: int((row.meta or {}).get('timer_at') or 0))
rows = rows[:limit]
claimed = []
for row in rows:
for row in result.scalars().all():
claim_id = str(uuid4())
row.meta = {
**(row.meta or {}),
@@ -192,6 +191,7 @@ async def claim_due_timers(now_ns: int, limit: int = 10) -> list[tuple[str, str]
'timer_started_at': now_ns,
'timer_claim_id': claim_id,
}
row.timer_at = None
row.updated_at = int(time.time())
claimed.append((row.id, claim_id))
await db.commit()
@@ -205,9 +205,8 @@ async def cancel_timers_for_chat(
async with get_async_db() as db:
result = await db.execute(
select(Chat)
.where(Chat.timer_at.isnot(None))
.where(Chat.user_id == user_id)
.where(Chat.meta['internal'].as_boolean().is_(True))
.where(Chat.meta['type'].as_string() == 'timer')
.where(Chat.meta['parent_chat_id'].as_string() == parent_chat_id)
.where(Chat.meta['status'].as_string() == 'pending')
)
@@ -222,6 +221,7 @@ async def cancel_timers_for_chat(
'timer_cancelled_at': now_ns,
'timer_cancelled_by': event,
}
row.timer_at = None
row.updated_at = int(time.time())
await db.commit()
@@ -289,12 +289,14 @@ async def execute_due_timer(app, timer_id: str, claim_id: str | None = None) ->
if await has_active_tasks(app.state.redis, parent_chat_id):
timer_row = await db.get(Chat, timer_id)
if timer_row:
timer_meta = timer_row.meta or {}
timer_row.meta = {
**(timer_row.meta or {}),
**timer_meta,
'status': 'pending',
'timer_claim_id': None,
'timer_started_at': None,
}
timer_row.timer_at = timer_meta.get('timer_at')
timer_row.updated_at = int(time.time())
await db.commit()
return