Enhancement: Add a database caching for improved performance (#9784)

---------

Co-authored-by: shamoon <4887959+shamoon@users.noreply.github.com>
This commit is contained in:
Antoine Mérino
2025-07-01 07:36:24 +02:00
committed by GitHub
parent 6b248ef140
commit 1671d49d44
7 changed files with 309 additions and 16 deletions

17
src/paperless/db_cache.py Normal file
View File

@@ -0,0 +1,17 @@
from cachalot.api import invalidate as cachalot_invalidate
from cachalot.utils import get_query_cache_key
from cachalot.utils import get_table_cache_key
PREFIX = "pngx_cachalot_"
def custom_get_query_cache_key(compiler):
return PREFIX + get_query_cache_key(compiler)
def custom_get_table_cache_key(db_alias, table):
return PREFIX + get_table_cache_key(db_alias, table)
def invalidate_db_cache():
return cachalot_invalidate(cache_alias="read-cache")

View File

@@ -433,6 +433,7 @@ STORAGES = {
_CELERY_REDIS_URL, _CHANNELS_REDIS_URL = _parse_redis_url(
os.getenv("PAPERLESS_REDIS", None),
)
_REDIS_KEY_PREFIX = os.getenv("PAPERLESS_REDIS_PREFIX", "")
TEMPLATES = [
{
@@ -458,7 +459,7 @@ CHANNEL_LAYERS = {
"hosts": [_CHANNELS_REDIS_URL],
"capacity": 2000, # default 100
"expiry": 15, # default 60
"prefix": os.getenv("PAPERLESS_REDIS_PREFIX", ""),
"prefix": _REDIS_KEY_PREFIX,
},
},
}
@@ -882,7 +883,7 @@ CELERY_SEND_TASK_SENT_EVENT = True
CELERY_BROKER_CONNECTION_RETRY = True
CELERY_BROKER_CONNECTION_RETRY_ON_STARTUP = True
CELERY_BROKER_TRANSPORT_OPTIONS = {
"global_keyprefix": os.getenv("PAPERLESS_REDIS_PREFIX", ""),
"global_keyprefix": _REDIS_KEY_PREFIX,
}
CELERY_TASK_TRACK_STARTED = True
@@ -903,22 +904,69 @@ CELERY_BEAT_SCHEDULE = _parse_beat_schedule()
# https://docs.celeryq.dev/en/stable/userguide/configuration.html#beat-schedule-filename
CELERY_BEAT_SCHEDULE_FILENAME = str(DATA_DIR / "celerybeat-schedule.db")
# django setting.
CACHES = {
"default": {
"BACKEND": os.environ.get(
"PAPERLESS_CACHE_BACKEND",
"django.core.cache.backends.redis.RedisCache",
),
"LOCATION": _CHANNELS_REDIS_URL,
"KEY_PREFIX": os.getenv("PAPERLESS_REDIS_PREFIX", ""),
},
}
if DEBUG and os.getenv("PAPERLESS_CACHE_BACKEND") is None:
CACHES["default"]["BACKEND"] = (
"django.core.cache.backends.locmem.LocMemCache" # pragma: no cover
# Cachalot: Database read cache.
def _parse_cachalot_settings():
global INSTALLED_APPS
ttl = __get_int("PAPERLESS_READ_CACHE_TTL", 3600)
ttl = min(ttl, 31536000) if ttl > 0 else 3600
_, redis_url = _parse_redis_url(
os.getenv("PAPERLESS_READ_CACHE_REDIS_URL", None),
)
result = {
"CACHALOT_CACHE": "read-cache",
"CACHALOT_ENABLED": __get_boolean(
"PAPERLESS_DB_READ_CACHE_ENABLED",
default="no",
),
"CACHALOT_FINAL_SQL_CHECK": True,
"CACHALOT_QUERY_KEYGEN": "paperless.db_cache.custom_get_query_cache_key",
"CACHALOT_TABLE_KEYGEN": "paperless.db_cache.custom_get_table_cache_key",
"CACHALOT_REDIS_URL": redis_url,
"CACHALOT_TIMEOUT": ttl,
}
if result["CACHALOT_ENABLED"]:
INSTALLED_APPS.append("cachalot")
return result
_cachalot_settings = _parse_cachalot_settings()
CACHALOT_ENABLED = _cachalot_settings["CACHALOT_ENABLED"]
CACHALOT_CACHE = _cachalot_settings["CACHALOT_CACHE"]
CACHALOT_TIMEOUT = _cachalot_settings["CACHALOT_TIMEOUT"]
CACHALOT_QUERY_KEYGEN = _cachalot_settings["CACHALOT_QUERY_KEYGEN"]
CACHALOT_TABLE_KEYGEN = _cachalot_settings["CACHALOT_TABLE_KEYGEN"]
CACHALOT_FINAL_SQL_CHECK = _cachalot_settings["CACHALOT_FINAL_SQL_CHECK"]
# Django default & Cachalot cache configuration
_CACHE_BACKEND = os.environ.get(
"PAPERLESS_CACHE_BACKEND",
"django.core.cache.backends.locmem.LocMemCache"
if DEBUG
else "django.core.cache.backends.redis.RedisCache",
)
def _parse_caches():
return {
"default": {
"BACKEND": _CACHE_BACKEND,
"LOCATION": _CHANNELS_REDIS_URL,
"KEY_PREFIX": _REDIS_KEY_PREFIX,
},
"read-cache": {
"BACKEND": _CACHE_BACKEND,
"LOCATION": _parse_cachalot_settings()["CACHALOT_REDIS_URL"],
"KEY_PREFIX": _REDIS_KEY_PREFIX,
},
}
CACHES = _parse_caches()
del _cachalot_settings
def default_threads_per_worker(task_workers) -> int:

View File

@@ -0,0 +1,162 @@
import os
import time
from unittest.mock import patch
import pytest
from cachalot.settings import cachalot_settings
from django.conf import settings
from django.db import connection
from django.test import override_settings
from django.test.utils import CaptureQueriesContext
from documents.models import Tag
from paperless.db_cache import invalidate_db_cache
from paperless.settings import _parse_cachalot_settings
from paperless.settings import _parse_caches
def test_all_redis_caches_have_same_custom_prefix(monkeypatch):
"""
Check that when setting a custom Redis prefix,
it is set for both the Django default cache and the read cache.
"""
from paperless import settings
monkeypatch.setattr(settings, "_REDIS_KEY_PREFIX", "test_a_custom_key_prefix")
caches = _parse_caches()
assert caches["read-cache"]["KEY_PREFIX"] == "test_a_custom_key_prefix"
assert caches["default"]["KEY_PREFIX"] == "test_a_custom_key_prefix"
class TestDbCacheSettings:
def test_cachalot_default_settings(self):
# Cachalot must be installed even if disabled,
# so the cache can be invalidated anytime
assert "cachalot" not in settings.INSTALLED_APPS
cachalot_settings = _parse_cachalot_settings()
caches = _parse_caches()
# Default settings
assert not cachalot_settings["CACHALOT_ENABLED"]
assert cachalot_settings["CACHALOT_TIMEOUT"] == 3600
assert caches["read-cache"]["KEY_PREFIX"] == ""
assert caches["read-cache"]["LOCATION"] == "redis://localhost:6379"
# Fixed settings
assert cachalot_settings["CACHALOT_CACHE"] == "read-cache"
assert (
cachalot_settings["CACHALOT_QUERY_KEYGEN"]
== "paperless.db_cache.custom_get_query_cache_key"
)
assert (
cachalot_settings["CACHALOT_TABLE_KEYGEN"]
== "paperless.db_cache.custom_get_table_cache_key"
)
assert cachalot_settings["CACHALOT_FINAL_SQL_CHECK"] is True
@patch.dict(
os.environ,
{
"PAPERLESS_DB_READ_CACHE_ENABLED": "true",
"PAPERLESS_READ_CACHE_REDIS_URL": "redis://localhost:6380/7",
"PAPERLESS_READ_CACHE_TTL": "7200",
},
)
def test_cachalot_custom_settings(self):
cachalot_settings = _parse_cachalot_settings()
assert "cachalot" in settings.INSTALLED_APPS
caches = _parse_caches()
# Modifiable settings
assert cachalot_settings["CACHALOT_ENABLED"]
assert cachalot_settings["CACHALOT_TIMEOUT"] == 7200
assert caches["read-cache"]["LOCATION"] == "redis://localhost:6380/7"
# Fixed settings
assert cachalot_settings["CACHALOT_CACHE"] == "read-cache"
assert (
cachalot_settings["CACHALOT_QUERY_KEYGEN"]
== "paperless.db_cache.custom_get_query_cache_key"
)
assert (
cachalot_settings["CACHALOT_TABLE_KEYGEN"]
== "paperless.db_cache.custom_get_table_cache_key"
)
assert cachalot_settings["CACHALOT_FINAL_SQL_CHECK"] is True
@pytest.mark.parametrize(
("env_var_ttl", "expected_cachalot_timeout"),
[
# 0 or less will be ignored, and the default TTL will be set
("0", 3600),
("-1", 3600),
("-500000", 3600),
# Any positive value will be set, for a maximum of one year
("1", 1),
("7524", 7524),
("99999999999999", 31536000),
],
)
def test_cachalot_ttl_parsing(
self,
env_var_ttl: int,
expected_cachalot_timeout: int,
):
with patch.dict(os.environ, {"PAPERLESS_READ_CACHE_TTL": f"{env_var_ttl}"}):
cachalot_timeout = _parse_cachalot_settings()["CACHALOT_TIMEOUT"]
assert cachalot_timeout == expected_cachalot_timeout
@override_settings(
CACHALOT_ENABLED=True,
CACHALOT_TIMEOUT=1,
)
@pytest.mark.django_db(transaction=True)
def test_cache_hit_when_enabled():
cachalot_settings.reload()
assert cachalot_settings.CACHALOT_ENABLED
assert cachalot_settings.CACHALOT_TIMEOUT == 1
assert settings.CACHALOT_TIMEOUT == 1
# Read a table to populate the cache
list(list(Tag.objects.values_list("id", flat=True)))
# Invalidate the cache then read the database, there should be DB hit
invalidate_db_cache()
with CaptureQueriesContext(connection) as ctx:
list(list(Tag.objects.values_list("id", flat=True)))
assert len(ctx)
# Doing the same request again should hit the cache, not the DB
with CaptureQueriesContext(connection) as ctx:
list(list(Tag.objects.values_list("id", flat=True)))
assert not len(ctx)
# Wait the end of TTL
# Redis expire accuracy should be between 0 and 1 ms
time.sleep(1.002)
# Read the DB again. The DB should be hit because the cache has expired
with CaptureQueriesContext(connection) as ctx:
list(list(Tag.objects.values_list("id", flat=True)))
assert len(ctx)
# Invalidate the cache at the end of test
invalidate_db_cache()
@pytest.mark.django_db(transaction=True)
def test_cache_is_disabled_by_default():
cachalot_settings.reload()
# Invalidate the cache just in case
invalidate_db_cache()
# Read the table multiple times: the DB should always be hit without cache
for _ in range(3):
with CaptureQueriesContext(connection) as ctx:
list(list(Tag.objects.values_list("id", flat=True)))
assert len(ctx)
# Invalidate the cache at the end of test
invalidate_db_cache()