fix: Performance and high memory usage of relation hint calculation

* Calculation of hint message when requested relation is not present in schema cache requires creation of a FuzzySet (to use fuzzy search to find candidate tables). For schemas with many tables it is costly.
This patch introduces dbTablesFuzzyIndex in SchemaCache to memoize the FuzzySet creation.

* Additionally, because of FuzzySet large memory requirements, this patch introduces a limit of 500 relations per schema, above which FuzzySet is not created and hint calculation disabled.
This commit is contained in:
Michał Kłeczek
2026-01-03 07:56:12 +08:00
committed by Steve Chavez
parent 9ec5b030ce
commit e592d568c6
7 changed files with 86 additions and 29 deletions
+1
View File
@@ -20,6 +20,7 @@ All notable changes to this project will be documented in this file. From versio
- Fix not returning `Content-Length` on empty HTTP `201` responses by @laurenceisla in #4518 - Fix not returning `Content-Length` on empty HTTP `201` responses by @laurenceisla in #4518
- Fix inaccurate Server-Timing header durations by @steve-chavez in #4522 - Fix inaccurate Server-Timing header durations by @steve-chavez in #4522
- Fix inaccurate "Schema cache queried" logs by @steve-chavez in #4522 - Fix inaccurate "Schema cache queried" logs by @steve-chavez in #4522
- Fix performance and high memory usage of relation hint calculation by @mkleczek in #4462 #4463
## [14.1] - 2025-11-05 ## [14.1] - 2025-11-05
+7 -9
View File
@@ -3,6 +3,7 @@ Module : PostgREST.Error
Description : PostgREST error HTTP responses Description : PostgREST error HTTP responses
-} -}
{-# OPTIONS_GHC -fno-warn-orphans #-} {-# OPTIONS_GHC -fno-warn-orphans #-}
{-# LANGUAGE NamedFieldPuns #-}
{-# LANGUAGE RecordWildCards #-} {-# LANGUAGE RecordWildCards #-}
module PostgREST.Error module PostgREST.Error
@@ -41,6 +42,7 @@ import Network.HTTP.Types.Header (Header)
import PostgREST.MediaType (MediaType (..)) import PostgREST.MediaType (MediaType (..))
import qualified PostgREST.MediaType as MediaType import qualified PostgREST.MediaType as MediaType
import PostgREST.SchemaCache (SchemaCache (SchemaCache, dbTablesFuzzyIndex))
import PostgREST.SchemaCache.Identifiers (QualifiedIdentifier (..), import PostgREST.SchemaCache.Identifiers (QualifiedIdentifier (..),
Schema) Schema)
import PostgREST.SchemaCache.Relationship (Cardinality (..), import PostgREST.SchemaCache.Relationship (Cardinality (..),
@@ -49,10 +51,8 @@ import PostgREST.SchemaCache.Relationship (Cardinality (..),
RelationshipsMap) RelationshipsMap)
import PostgREST.SchemaCache.Routine (Routine (..), import PostgREST.SchemaCache.Routine (Routine (..),
RoutineParam (..)) RoutineParam (..))
import PostgREST.SchemaCache.Table (Table (..))
import Protolude import Protolude
class (ErrorBody a, JSON.ToJSON a) => PgrstError a where class (ErrorBody a, JSON.ToJSON a) => PgrstError a where
status :: a -> HTTP.Status status :: a -> HTTP.Status
headers :: a -> [Header] headers :: a -> [Header]
@@ -250,7 +250,7 @@ data SchemaCacheError
| NoRelBetween Text Text (Maybe Text) Text RelationshipsMap | NoRelBetween Text Text (Maybe Text) Text RelationshipsMap
| NoRpc Text Text [Text] MediaType Bool [QualifiedIdentifier] [Routine] | NoRpc Text Text [Text] MediaType Bool [QualifiedIdentifier] [Routine]
| ColumnNotFound Text Text | ColumnNotFound Text Text
| TableNotFound Text Text [Table] | TableNotFound Text Text SchemaCache
deriving Show deriving Show
instance PgrstError SchemaCacheError where instance PgrstError SchemaCacheError where
@@ -313,7 +313,7 @@ instance ErrorBody SchemaCacheError where
where where
onlySingleParams = isInvPost && contentType `elem` [MTTextPlain, MTTextXML, MTOctetStream] onlySingleParams = isInvPost && contentType `elem` [MTTextPlain, MTTextXML, MTOctetStream]
hint (AmbiguousRpc _) = Just "Try renaming the parameters or the function itself in the database so function overloading can be resolved" hint (AmbiguousRpc _) = Just "Try renaming the parameters or the function itself in the database so function overloading can be resolved"
hint (TableNotFound schemaName relName tbls) = JSON.String <$> tableNotFoundHint schemaName relName tbls hint (TableNotFound schemaName relName schemaCache) = JSON.String <$> tableNotFoundHint schemaName relName schemaCache
hint _ = Nothing hint _ = Nothing
@@ -428,13 +428,11 @@ noRpcHint schema procName params allProcs overloadedProcs =
-- | -- |
-- Do a fuzzy search in all tables in the same schema and return closest result -- Do a fuzzy search in all tables in the same schema and return closest result
tableNotFoundHint :: Text -> Text -> [Table] -> Maybe Text tableNotFoundHint :: Text -> Text -> SchemaCache -> Maybe Text
tableNotFoundHint schema tblName tblList tableNotFoundHint schema tblName SchemaCache{dbTablesFuzzyIndex}
= fmap (\tbl -> "Perhaps you meant the table '" <> schema <> "." <> tbl <> "'") perhapsTable = fmap (\tbl -> "Perhaps you meant the table '" <> schema <> "." <> tbl <> "'") perhapsTable
where where
perhapsTable = Fuzzy.getOne fuzzyTableSet tblName perhapsTable = (`Fuzzy.getOne` tblName) =<< HM.lookup schema dbTablesFuzzyIndex
fuzzyTableSet = Fuzzy.fromList [ tableName tbl | tbl <- tblList, tableSchema tbl == schema]
compressedRel :: Relationship -> JSON.Value compressedRel :: Relationship -> JSON.Value
-- An ambiguousness error cannot happen for computed relationships TODO refactor so this mempty is not needed -- An ambiguousness error cannot happen for computed relationships TODO refactor so this mempty is not needed
+6 -6
View File
@@ -170,7 +170,7 @@ dbActionPlan dbAct conf apiReq sCache = case dbAct of
wrappedReadPlan :: QualifiedIdentifier -> AppConfig -> SchemaCache -> ApiRequest -> Bool -> Either Error CrudPlan wrappedReadPlan :: QualifiedIdentifier -> AppConfig -> SchemaCache -> ApiRequest -> Bool -> Either Error CrudPlan
wrappedReadPlan identifier conf sCache apiRequest@ApiRequest{iPreferences=Preferences{..},..} headersOnly = do wrappedReadPlan identifier conf sCache apiRequest@ApiRequest{iPreferences=Preferences{..},..} headersOnly = do
qi <- findTable identifier (dbTables sCache) qi <- findTable identifier sCache
rPlan <- readPlan qi conf sCache apiRequest rPlan <- readPlan qi conf sCache apiRequest
(handler, mediaType) <- mapLeft ApiRequestError $ negotiateContent conf apiRequest qi iAcceptMediaType (dbMediaHandlers sCache) (hasDefaultSelect rPlan) (handler, mediaType) <- mapLeft ApiRequestError $ negotiateContent conf apiRequest qi iAcceptMediaType (dbMediaHandlers sCache) (hasDefaultSelect rPlan)
if not (null invalidPrefs) && preferHandling == Just Strict then Left $ ApiRequestError $ InvalidPreferences invalidPrefs else Right () if not (null invalidPrefs) && preferHandling == Just Strict then Left $ ApiRequestError $ InvalidPreferences invalidPrefs else Right ()
@@ -178,7 +178,7 @@ wrappedReadPlan identifier conf sCache apiRequest@ApiRequest{iPreferences=Prefe
mutateReadPlan :: Mutation -> ApiRequest -> QualifiedIdentifier -> AppConfig -> SchemaCache -> Either Error CrudPlan mutateReadPlan :: Mutation -> ApiRequest -> QualifiedIdentifier -> AppConfig -> SchemaCache -> Either Error CrudPlan
mutateReadPlan mutation apiRequest@ApiRequest{iPreferences=Preferences{..},..} identifier conf sCache = do mutateReadPlan mutation apiRequest@ApiRequest{iPreferences=Preferences{..},..} identifier conf sCache = do
qi <- findTable identifier (dbTables sCache) qi <- findTable identifier sCache
rPlan <- readPlan qi conf sCache apiRequest rPlan <- readPlan qi conf sCache apiRequest
mPlan <- mutatePlan mutation qi apiRequest sCache rPlan mPlan <- mutatePlan mutation qi apiRequest sCache rPlan
if not (null invalidPrefs) && preferHandling == Just Strict then Left $ ApiRequestError $ InvalidPreferences invalidPrefs else Right () if not (null invalidPrefs) && preferHandling == Just Strict then Left $ ApiRequestError $ InvalidPreferences invalidPrefs else Right ()
@@ -810,10 +810,10 @@ validateAggFunctions aggFunctionsAllowed (Node rp@ReadPlan {select} forest)
| otherwise = Node rp <$> traverse (validateAggFunctions aggFunctionsAllowed) forest | otherwise = Node rp <$> traverse (validateAggFunctions aggFunctionsAllowed) forest
-- | Lookup table in the schema cache before creating read plan -- | Lookup table in the schema cache before creating read plan
findTable :: QualifiedIdentifier -> TablesMap -> Either Error QualifiedIdentifier findTable :: QualifiedIdentifier -> SchemaCache -> Either Error QualifiedIdentifier
findTable qi@QualifiedIdentifier{..} tableMap = findTable qi@QualifiedIdentifier{..} sc@SchemaCache{dbTables} =
case HM.lookup qi tableMap of case HM.lookup qi dbTables of
Nothing -> Left $ SchemaCacheErr $ TableNotFound qiSchema qiName (HM.elems tableMap) Nothing -> Left $ SchemaCacheErr $ TableNotFound qiSchema qiName sc
Just _ -> Right qi Just _ -> Right qi
addFilters :: ResolverContext -> ApiRequest -> ReadPlanTree -> Either Error ReadPlanTree addFilters :: ResolverContext -> ApiRequest -> ReadPlanTree -> Either Error ReadPlanTree
+2 -2
View File
@@ -213,10 +213,10 @@ actionResponse (MaybeDbResult InspectPlan{ipHdrsOnly=headersOnly} body) _ versio
in in
Right $ PgrstResponse HTTP.status200 (MediaType.toContentType MTOpenAPI : cLHeader ++ maybeToList (profileHeader schema negotiatedByProfile)) rsBody Right $ PgrstResponse HTTP.status200 (MediaType.toContentType MTOpenAPI : cLHeader ++ maybeToList (profileHeader schema negotiatedByProfile)) rsBody
actionResponse (NoDbResult (RelInfoPlan qi@QualifiedIdentifier{..})) _ _ _ SchemaCache{dbTables} _ _ = actionResponse (NoDbResult (RelInfoPlan qi@QualifiedIdentifier{..})) _ _ _ sc@SchemaCache{dbTables} _ _ =
case HM.lookup qi dbTables of case HM.lookup qi dbTables of
Just tbl -> respondInfo $ allowH tbl Just tbl -> respondInfo $ allowH tbl
Nothing -> Left $ Error.SchemaCacheErr $ Error.TableNotFound qiSchema qiName (HM.elems dbTables) Nothing -> Left $ Error.SchemaCacheErr $ Error.TableNotFound qiSchema qiName sc
where where
allowH table = allowH table =
let hasPK = not . null $ tablePKCols table in let hasPK = not . null $ tablePKCols table in
+28 -12
View File
@@ -20,6 +20,7 @@ These queries are executed once at startup or when PostgREST is reloaded.
module PostgREST.SchemaCache module PostgREST.SchemaCache
( SchemaCache(..) ( SchemaCache(..)
, TablesFuzzyIndex
, querySchemaCache , querySchemaCache
, showSummary , showSummary
, decodeFuncs , decodeFuncs
@@ -66,21 +67,28 @@ import PostgREST.SchemaCache.Table (Column (..), ColumnMap,
import qualified PostgREST.MediaType as MediaType import qualified PostgREST.MediaType as MediaType
import Control.Arrow ((&&&)) import Control.Arrow ((&&&))
import Protolude import qualified Data.FuzzySet as Fuzzy
import System.IO.Unsafe (unsafePerformIO) import Protolude
import System.IO.Unsafe (unsafePerformIO)
type TablesFuzzyIndex = HM.HashMap Schema Fuzzy.FuzzySet
data SchemaCache = SchemaCache data SchemaCache = SchemaCache
{ dbTables :: TablesMap { dbTables :: TablesMap
, dbRelationships :: RelationshipsMap , dbRelationships :: RelationshipsMap
, dbRoutines :: RoutineMap , dbRoutines :: RoutineMap
, dbRepresentations :: RepresentationsMap , dbRepresentations :: RepresentationsMap
, dbMediaHandlers :: MediaHandlerMap , dbMediaHandlers :: MediaHandlerMap
, dbTimezones :: TimezoneNames , dbTimezones :: TimezoneNames
} -- Memoized fuzzy index of table names per schema to support approximate matching
-- Since index construction can be expensive, we build it once and store in the SchemaCache
-- Haskell lazy evaluation ensures it's only built on first use and memoized afterwards
, dbTablesFuzzyIndex :: TablesFuzzyIndex
} deriving (Show)
instance JSON.ToJSON SchemaCache where instance JSON.ToJSON SchemaCache where
toJSON (SchemaCache tabs rels routs reps hdlers tzs) = JSON.object [ toJSON (SchemaCache tabs rels routs reps hdlers tzs _) = JSON.object [
"dbTables" .= JSON.toJSON tabs "dbTables" .= JSON.toJSON tabs
, "dbRelationships" .= JSON.toJSON rels , "dbRelationships" .= JSON.toJSON rels
, "dbRoutines" .= JSON.toJSON routs , "dbRoutines" .= JSON.toJSON routs
@@ -90,7 +98,7 @@ instance JSON.ToJSON SchemaCache where
] ]
showSummary :: SchemaCache -> Text showSummary :: SchemaCache -> Text
showSummary (SchemaCache tbls rels routs reps mediaHdlrs tzs) = showSummary (SchemaCache tbls rels routs reps mediaHdlrs tzs _) =
T.intercalate ", " T.intercalate ", "
[ show (HM.size tbls) <> " Relations" [ show (HM.size tbls) <> " Relations"
, show (HM.size rels) <> " Relationships" , show (HM.size rels) <> " Relationships"
@@ -138,6 +146,8 @@ data KeyDep
-- | A SQL query that can be executed independently -- | A SQL query that can be executed independently
type SqlQuery = ByteString type SqlQuery = ByteString
maxDbTablesForFuzzySearch :: Int
maxDbTablesForFuzzySearch = 500
querySchemaCache :: AppConfig -> SQL.Transaction SchemaCache querySchemaCache :: AppConfig -> SQL.Transaction SchemaCache
querySchemaCache conf@AppConfig{..} = do querySchemaCache conf@AppConfig{..} = do
@@ -166,6 +176,11 @@ querySchemaCache conf@AppConfig{..} = do
, dbRepresentations = reps , dbRepresentations = reps
, dbMediaHandlers = HM.union mHdlers initialMediaHandlers -- the custom handlers will override the initial ones , dbMediaHandlers = HM.union mHdlers initialMediaHandlers -- the custom handlers will override the initial ones
, dbTimezones = tzones , dbTimezones = tzones
, dbTablesFuzzyIndex =
-- Only build fuzzy index for schemas with a reasonable number of tables
-- Fuzzy.FuzzySet is memory heavy we just don't use it for large schemas
Fuzzy.fromList <$> HM.filter ((< maxDbTablesForFuzzySearch) . length) (HM.fromListWith (<>) ((qiSchema &&& pure . qiName) <$> HM.keys tabsWViewsPks))
} }
where where
schemas = toList configDbSchemas schemas = toList configDbSchemas
@@ -203,6 +218,7 @@ removeInternal schemas dbStruct =
, dbRepresentations = dbRepresentations dbStruct -- no need to filter, not directly exposed through the API , dbRepresentations = dbRepresentations dbStruct -- no need to filter, not directly exposed through the API
, dbMediaHandlers = dbMediaHandlers dbStruct , dbMediaHandlers = dbMediaHandlers dbStruct
, dbTimezones = dbTimezones dbStruct , dbTimezones = dbTimezones dbStruct
, dbTablesFuzzyIndex = dbTablesFuzzyIndex dbStruct
} }
where where
hasInternalJunction ComputedRelationship{} = False hasInternalJunction ComputedRelationship{} = False
+22
View File
@@ -11375,12 +11375,34 @@ ALTER TABLE ONLY apflora.zielber
ALTER TABLE apflora."user" ENABLE ROW LEVEL SECURITY; ALTER TABLE apflora."user" ENABLE ROW LEVEL SECURITY;
CREATE SCHEMA fuzzysearch;
-- Create many tables to test fuzzy string search
-- computing hints for non existing tables
DO
$$
DECLARE
r record;
BEGIN
FOR r IN
SELECT
format('CREATE TABLE fuzzysearch.unknown_table_%s ()', n) AS ct
FROM
generate_series(1, 499) n
LOOP
EXECUTE r.ct;
END LOOP;
END
$$;
DROP ROLE IF EXISTS postgrest_test_anonymous; DROP ROLE IF EXISTS postgrest_test_anonymous;
CREATE ROLE postgrest_test_anonymous; CREATE ROLE postgrest_test_anonymous;
GRANT postgrest_test_anonymous TO :PGUSER; GRANT postgrest_test_anonymous TO :PGUSER;
GRANT USAGE ON SCHEMA apflora TO postgrest_test_anonymous; GRANT USAGE ON SCHEMA apflora TO postgrest_test_anonymous;
GRANT USAGE ON SCHEMA fuzzysearch TO postgrest_test_anonymous;
GRANT ALL PRIVILEGES ON ALL TABLES IN SCHEMA apflora GRANT ALL PRIVILEGES ON ALL TABLES IN SCHEMA apflora
TO postgrest_test_anonymous; TO postgrest_test_anonymous;
+20
View File
@@ -70,3 +70,23 @@ def test_should_not_fail_with_stack_overflow(defaultenv):
assert response.status_code == 404 assert response.status_code == 404
data = response.json() data = response.json()
assert data["code"] == "PGRST205" assert data["code"] == "PGRST205"
def test_second_request_for_non_existent_table_should_be_quick(defaultenv):
"requesting a non-existent relationship should be quick after the fuzzy search index is loaded (2nd request)"
env = {
**defaultenv,
"PGRST_DB_SCHEMAS": "fuzzysearch",
"PGRST_DB_POOL": "2",
"PGRST_DB_ANON_ROLE": "postgrest_test_anonymous",
}
with run(env=env, wait_max_seconds=30) as postgrest:
response = postgrest.session.get("/unknown-table")
assert response.status_code == 404
data = response.json()
assert data["code"] == "PGRST205"
first_duration = response.elapsed.total_seconds()
response = postgrest.session.get("/unknown-table")
assert response.elapsed.total_seconds() < first_duration / 10