FASE 7b: DB Performance — Pooling, Stock Summary, N+1 fix

Cambios implementados:

1. Connection pooling (tenant_db.py):
   - psycopg2.pool.ThreadedConnectionPool para master y tenants
   - Wrapper _PooledConnection que devuelve al pool en .close()
   - Cero cambios en blueprints (backward compatible)

2. Tabla inventory_stock_summary + triggers (v3.2):
   - O(1) stock lookup en vez de SUM() sobre historial completo
   - Trigger AFTER INSERT en inventory_operations recalcula stock
   - Poblada inicialmente en ambos tenants
   - Refactor en 6 archivos de servicios para usar la nueva tabla

3. Fix N+1 en process_sale (pos_engine.py):
   - Precarga retail_price en bulk query FOR UPDATE
   - Elimina SELECT individual por item en loop

4. Índices críticos:
   - idx_parts_name_part + pattern_ops (master)
   - idx_inv_ops_inventory_branch_created (tenants)
   - idx_wi_part_stock_positive (master, ya existía desde Fase 1)

Tests: 73/73 pasando (compat + fase3 + fase5 + fase6)
Migración: v3.2_db_performance.sql
This commit is contained in:
2026-04-27 07:34:31 +00:00
parent 175dda6212
commit e3c85fd647
9 changed files with 173 additions and 40 deletions

View File

@@ -32,6 +32,7 @@ MIGRATIONS = {
'v2.9': 'v2.9_logistics.sql',
'v3.0': 'v3.0_public_api.sql',
'v3.1': 'v3.1_inventory_vehicle_compat.sql',
'v3.2': 'v3.2_db_performance.sql',
}

View File

@@ -0,0 +1,62 @@
-- FASE 2: DB Performance Optimizations
-- Applies to: master DB (nexus_autoparts) and all tenant DBs
-- ─── MASTER DB ─────────────────────────────────
-- Index for parts name lookups (used by get_part_types, shop supplies)
CREATE INDEX IF NOT EXISTS idx_parts_name_part ON parts(name_part);
CREATE INDEX IF NOT EXISTS idx_parts_name_part_pattern ON parts(name_part text_pattern_ops);
-- Partial index for warehouse_inventory stock lookups (already created in Fase 1,
-- kept here for completeness on fresh installs)
CREATE INDEX IF NOT EXISTS idx_wi_part_stock_positive
ON warehouse_inventory(part_id) WHERE stock_quantity > 0;
-- ─── TENANT DB (run on each tenant) ────────────
-- O(1) stock summary table with trigger-based updates
CREATE TABLE IF NOT EXISTS inventory_stock_summary (
inventory_id INT PRIMARY KEY REFERENCES inventory(id) ON DELETE CASCADE,
branch_id INT REFERENCES branches(id),
stock INT NOT NULL DEFAULT 0,
last_updated TIMESTAMPTZ DEFAULT NOW()
);
CREATE INDEX IF NOT EXISTS idx_iss_branch ON inventory_stock_summary(branch_id);
-- Trigger function: recalculates stock for an item after every operation insert
CREATE OR REPLACE FUNCTION update_stock_summary()
RETURNS TRIGGER AS $$
BEGIN
INSERT INTO inventory_stock_summary (inventory_id, branch_id, stock)
SELECT i.id, i.branch_id, COALESCE(SUM(io.quantity), 0)
FROM inventory i
LEFT JOIN inventory_operations io ON io.inventory_id = i.id
WHERE i.id = NEW.inventory_id
GROUP BY i.id, i.branch_id
ON CONFLICT (inventory_id) DO UPDATE SET
stock = EXCLUDED.stock,
branch_id = EXCLUDED.branch_id,
last_updated = NOW();
RETURN NEW;
END;
$$ LANGUAGE plpgsql;
DROP TRIGGER IF EXISTS trg_update_stock_summary ON inventory_operations;
CREATE TRIGGER trg_update_stock_summary
AFTER INSERT ON inventory_operations
FOR EACH ROW EXECUTE FUNCTION update_stock_summary();
-- Composite index for inventory_operations (used by stock + history queries)
CREATE INDEX IF NOT EXISTS idx_inv_ops_inventory_branch_created
ON inventory_operations(inventory_id, branch_id, created_at DESC);
-- Initial population of summary table (idempotent)
INSERT INTO inventory_stock_summary (inventory_id, branch_id, stock)
SELECT i.id, i.branch_id, COALESCE(SUM(io.quantity), 0)
FROM inventory i
LEFT JOIN inventory_operations io ON io.inventory_id = i.id
GROUP BY i.id, i.branch_id
ON CONFLICT (inventory_id) DO UPDATE SET
stock = EXCLUDED.stock,
branch_id = EXCLUDED.branch_id,
last_updated = NOW();

View File

@@ -140,7 +140,7 @@ def get_inventory_context(tenant_conn, branch_id=None):
cur.execute(f"""
SELECT COUNT(*) FROM inventory i
WHERE {where}
AND COALESCE((SELECT SUM(quantity) FROM inventory_operations WHERE inventory_id = i.id), 0) <= 3
AND COALESCE((SELECT stock FROM inventory_stock_summary WHERE inventory_id = i.id), 0) <= 3
""", params)
low_stock = cur.fetchone()[0] or 0

View File

@@ -1411,12 +1411,11 @@ def _get_local_stock_bulk(tenant_conn, branch_id, oem_numbers, catalog_part_ids)
cur.execute(f"""
SELECT i.id, i.part_number, i.catalog_part_id,
i.price_1, i.price_2, i.price_3, i.cost, i.tax_rate,
COALESCE(SUM(io.quantity), 0) AS stock,
COALESCE(s.stock, 0) AS stock,
i.image_url
FROM inventory i
LEFT JOIN inventory_operations io ON io.inventory_id = i.id
LEFT JOIN inventory_stock_summary s ON s.inventory_id = i.id
WHERE ({where}) AND i.is_active = true{branch_filter}
GROUP BY i.id
""", params)
result = {}
@@ -1453,13 +1452,12 @@ def _get_local_stock_single(tenant_conn, branch_id, oem_part_number, catalog_par
cur.execute(f"""
SELECT i.id, i.price_1, i.price_2, i.price_3, i.cost, i.tax_rate,
i.location, i.unit, i.barcode,
COALESCE(SUM(io.quantity), 0) AS stock,
COALESCE(s.stock, 0) AS stock,
i.image_url
FROM inventory i
LEFT JOIN inventory_operations io ON io.inventory_id = i.id
LEFT JOIN inventory_stock_summary s ON s.inventory_id = i.id
WHERE (i.part_number = %s OR i.catalog_part_id = %s)
AND i.is_active = true{branch_filter}
GROUP BY i.id
LIMIT 1
""", params)

View File

@@ -25,15 +25,33 @@ def _safe_g(attr, default=None):
def get_stock(conn, inventory_id, branch_id=None):
"""Get current stock for an inventory item. Optionally filter by branch.
Uses Redis cache first, falls back to PostgreSQL SUM query.
Uses Redis cache first, then inventory_stock_summary, falls back to
PostgreSQL SUM query.
"""
# Try Redis first
cached = get_cached_stock(inventory_id, branch_id)
if cached is not None:
return cached
# Fallback to PostgreSQL
# Use inventory_stock_summary (O(1) lookup)
cur = conn.cursor()
if branch_id:
cur.execute(
"SELECT stock FROM inventory_stock_summary WHERE inventory_id = %s AND branch_id = %s",
(inventory_id, branch_id)
)
else:
cur.execute(
"SELECT stock FROM inventory_stock_summary WHERE inventory_id = %s",
(inventory_id,)
)
row = cur.fetchone()
if row is not None:
cur.close()
set_cached_stock(inventory_id, row[0], branch_id)
return row[0]
# Fallback to PostgreSQL SUM (legacy, should not reach here if trigger works)
if branch_id:
cur.execute(
"SELECT COALESCE(SUM(quantity), 0) FROM inventory_operations WHERE inventory_id = %s AND branch_id = %s",
@@ -55,21 +73,17 @@ def get_stock(conn, inventory_id, branch_id=None):
def get_stock_bulk(conn, branch_id=None):
"""Get stock for all items. Returns dict {inventory_id: stock_quantity}.
Uses PostgreSQL directly (bulk operation, Redis wouldn't help much here
unless we pre-populated all keys).
Uses inventory_stock_summary for O(1) bulk lookup.
"""
cur = conn.cursor()
if branch_id:
cur.execute("""
SELECT inventory_id, COALESCE(SUM(quantity), 0)
FROM inventory_operations WHERE branch_id = %s
GROUP BY inventory_id
SELECT inventory_id, stock
FROM inventory_stock_summary WHERE branch_id = %s
""", (branch_id,))
else:
cur.execute("""
SELECT inventory_id, COALESCE(SUM(quantity), 0)
FROM inventory_operations
GROUP BY inventory_id
SELECT inventory_id, stock FROM inventory_stock_summary
""")
stock_map = {r[0]: r[1] for r in cur.fetchall()}
cur.close()

View File

@@ -194,11 +194,7 @@ def get_inventory_by_vehicle(tenant_conn, master_conn, mye_id, branch_id=None):
i.image_url, i.description, COALESCE(s.stock, 0) as stock
FROM inventory i
JOIN inventory_vehicle_compat ivc ON ivc.inventory_id = i.id
LEFT JOIN (
SELECT inventory_id, SUM(quantity) as stock
FROM inventory_operations
GROUP BY inventory_id
) s ON s.inventory_id = i.id
LEFT JOIN inventory_stock_summary s ON s.inventory_id = i.id
WHERE ivc.model_year_engine_id = %s {branch_filter}
AND i.is_active = true
ORDER BY i.name

View File

@@ -93,11 +93,7 @@ def get_local_inventory(tenant_conn, query: str = None, limit: int = 50) -> list
COALESCE(s.stock, 0) AS stock,
i.unit, i.catalog_part_id
FROM inventory i
LEFT JOIN (
SELECT inventory_id, SUM(quantity) AS stock
FROM inventory_operations
GROUP BY inventory_id
) s ON s.inventory_id = i.id
LEFT JOIN inventory_stock_summary s ON s.inventory_id = i.id
WHERE {where}
ORDER BY i.name
LIMIT %s

View File

@@ -224,7 +224,7 @@ def process_sale(conn, sale_data):
# Lock inventory rows to prevent race conditions on concurrent sales
cur.execute("""
SELECT id, part_number, name, cost, price_1, price_2, price_3,
tax_rate, branch_id
tax_rate, branch_id, retail_price
FROM inventory
WHERE id = ANY(%s) AND is_active = true
ORDER BY id
@@ -327,10 +327,9 @@ def process_sale(conn, sale_data):
# Create sale items (batch insert) and deduct inventory
sale_items_data = []
for item in totals['items']:
# Fetch retail_price for savings calculation
cur.execute("SELECT retail_price FROM inventory WHERE id = %s", (item['inventory_id'],))
rp_row = cur.fetchone()
retail_price = rp_row[0] if rp_row else None
# retail_price from preloaded bulk query (index 9)
inv = inv_rows.get(item['inventory_id'])
retail_price = inv[9] if inv else None
sale_items_data.append((
sale_id, item['inventory_id'], item['part_number'], item['name'],

View File

@@ -1,20 +1,85 @@
# /home/Autopartes/pos/tenant_db.py
"""Tenant DB connection manager. Gets a psycopg2 connection for a specific tenant."""
"""Tenant DB connection manager with pooling.
Uses psycopg2.pool.ThreadedConnectionPool for both master and tenant DBs.
Connections are returned to the pool on .close() via a thin wrapper —
this keeps the rest of the codebase unchanged.
"""
import psycopg2
from psycopg2 import pool
from config import MASTER_DB_URL, TENANT_DB_URL_TEMPLATE
# ─── Pools ─────────────────────────────────────
_master_pool = None
_tenant_pools = {}
def _get_master_pool():
"""Lazy-initialize master DB connection pool."""
global _master_pool
if _master_pool is None:
_master_pool = pool.ThreadedConnectionPool(
minconn=2, maxconn=10, dsn=MASTER_DB_URL
)
return _master_pool
def _get_tenant_pool(db_name):
"""Lazy-initialize tenant DB connection pool by db_name."""
global _tenant_pools
if db_name not in _tenant_pools:
dsn = TENANT_DB_URL_TEMPLATE.format(db_name=db_name)
_tenant_pools[db_name] = pool.ThreadedConnectionPool(
minconn=2, maxconn=10, dsn=dsn
)
return _tenant_pools[db_name]
class _PooledConnection:
"""Thin wrapper that delegates all attribute access to the real
psycopg2 connection, but returns it to the pool on .close().
"""
__slots__ = ('_conn', '_pool')
def __init__(self, conn, db_pool):
self._conn = conn
self._pool = db_pool
def __getattr__(self, name):
return getattr(self._conn, name)
def close(self):
try:
self._pool.putconn(self._conn)
except Exception:
# If pool is already closed, fall back to real close
self._conn.close()
def __enter__(self):
return self
def __exit__(self, exc_type, exc_val, exc_tb):
self.close()
# ─── Public API ────────────────────────────────
def get_master_conn():
"""Get connection to nexus_master DB."""
return psycopg2.connect(MASTER_DB_URL)
"""Get a pooled connection to the master DB."""
p = _get_master_pool()
return _PooledConnection(p.getconn(), p)
def get_tenant_conn(tenant_id):
"""Get connection to a tenant's DB by looking up db_name in nexus_master."""
"""Get a pooled connection to a tenant's DB."""
master = get_master_conn()
cur = master.cursor()
cur.execute("SELECT db_name FROM tenants WHERE id = %s AND is_active = true", (tenant_id,))
cur.execute(
"SELECT db_name FROM tenants WHERE id = %s AND is_active = true",
(tenant_id,)
)
row = cur.fetchone()
cur.close()
master.close()
@@ -23,9 +88,11 @@ def get_tenant_conn(tenant_id):
raise ValueError(f"Tenant {tenant_id} not found or inactive")
db_name = row[0]
return psycopg2.connect(TENANT_DB_URL_TEMPLATE.format(db_name=db_name))
p = _get_tenant_pool(db_name)
return _PooledConnection(p.getconn(), p)
def get_tenant_conn_by_dbname(db_name):
"""Get connection to a tenant DB directly by name."""
return psycopg2.connect(TENANT_DB_URL_TEMPLATE.format(db_name=db_name))
"""Get a pooled connection to a tenant DB directly by name."""
p = _get_tenant_pool(db_name)
return _PooledConnection(p.getconn(), p)