graphql-engine/server/src-lib/Hasura/RQL/DDL/Schema/Cache/Common.hs
Auke Booij 67b922bac1 Avoid GraphQL schema rebuild when changing irrelevant Metadata
This increases the speed of `create_query_collection` and `add_collection_to_allowlist` by a factor ~~10~~ 65, by caching the in-memory GraphQL schema. This speedup also applies more broadly to Metadata changes relating to:
- allowlists
- query collections
- cron triggers
- REST endpoints
- API limits
- metrics config
- GraphQL introspection options
- TLS allow lists
- OpenTelemetry

When is construction of the in-memory GraphQL schema cached between Metadata operations?

Before this PR, **never**! It's rebuilt fully, for every role, on every Metadata operation.

However, there are many Metadata operations that don't influence the GraphQL schema. So we should be caching its construction.

The `Hasura.Incremental` framework allows us to cache such constructions: whenever we have an arrow `Rule m a b`, where `a` is the input to the arrow and `b` the output, we can use the `Inc.cache` combinator to obtain a new arrow which is only re-executed when the input `a` changes in a material way. To test this, `a` needs an `Eq` instance. (Before hasura/graphql-engine-mono#6877, this was a `Cacheable` type class which has now been removed.)

We can't simply apply `Inc.cache` to the "Steps 3 and 4" in `buildSchemaCacheRule`, because the inputs (components of `BuildOutputs` such as `SourceCache`) don't have an `Eq` instance.

So the changes to `buildSchemaCacheRule` restructure the code so that the input to "Step 1", namely the Metadata, can be used as a caching key instead, so that `Inc.cache` can be applied to the whole sequence of steps.

That works to cache construction of the GraphQL schema, but it means that now only those Metadata operations that _don't_ influence any of the products of steps 1-4 can use a cached build of the GraphQL schema. The most important intermediate product is `BuildOutputs`. So now the exercise becomes to minimize the amount of stuff stored in `BuildOutputs`, so that as many Metadata operations as possible can be handled outside of the codepath that produces a GraphQL schema.

Per hasura/graphql-engine-mono#6609, the `BuildOutputs` structure is too big, and stores things unnecessarily. Refer to the PR description there for reasoning - the same logic applies to this PR, and simply goes a few steps further. In doing so, it can benefit from hasura/graphql-engine-mono#6765, which allows us to verify at compile time that certain Schema Cache building steps _don't_ generate "Metadata dependencies". If a certain Metadata dependency is never generated, we don't need to handle that case in `deleteMetadataObject`. Thus such intermediate products don't need to be passed through `resolveDependencies`, and thus they don't need to be stored in `BuildOutputs`, and thus their rebuild won't trigger a GraphQL schema rebuild.

PR-URL: https://github.com/hasura/graphql-engine-mono/pull/6613
GitOrigin-RevId: 27d2e69d3461bd4c32f08febef9995c0369fab3a
2022-11-23 07:54:01 +00:00

347 lines
12 KiB
Haskell
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

{-# LANGUAGE Arrows #-}
{-# LANGUAGE TemplateHaskell #-}
{-# LANGUAGE UndecidableInstances #-}
-- | Types/functions shared between modules that implement "Hasura.RQL.DDL.Schema.Cache". Other
-- modules should not import this module directly.
module Hasura.RQL.DDL.Schema.Cache.Common
( ApolloFederationConfig (..),
ApolloFederationVersion (..),
BackendInvalidationKeysWrapper (..),
BuildOutputs (..),
CacheBuild,
CacheBuildParams (CacheBuildParams),
InvalidationKeys (..),
ikMetadata,
ikRemoteSchemas,
ikSources,
ikBackends,
NonColumnTableInputs (..),
RebuildableSchemaCache (RebuildableSchemaCache, lastBuiltSchemaCache),
TableBuildInput (TableBuildInput, _tbiName),
TablePermissionInputs (..),
addTableContext,
bindErrorA,
boActions,
boCustomTypes,
boBackendCache,
boRemoteSchemas,
boRoles,
boSources,
buildInfoMap,
buildInfoMapPreservingMetadata,
initialInvalidationKeys,
invalidateKeys,
mkTableInputs,
runCacheBuild,
runCacheBuildM,
withRecordDependencies,
)
where
import Control.Arrow.Extended
import Control.Arrow.Interpret
import Control.Lens
import Control.Monad.Trans.Control (MonadBaseControl)
import Control.Monad.Unique
import Data.HashMap.Strict.Extended qualified as M
import Data.HashMap.Strict.InsOrd qualified as OMap
import Data.Sequence qualified as Seq
import Data.Text.Extended
import Hasura.Base.Error
import Hasura.Incremental qualified as Inc
import Hasura.Prelude
import Hasura.RQL.Types.Backend
import Hasura.RQL.Types.Common
import Hasura.RQL.Types.CustomTypes
import Hasura.RQL.Types.Metadata
import Hasura.RQL.Types.Metadata.Backend (BackendMetadata (..))
import Hasura.RQL.Types.Metadata.Instances ()
import Hasura.RQL.Types.Metadata.Object
import Hasura.RQL.Types.Permission
import Hasura.RQL.Types.Relationships.Local
import Hasura.RQL.Types.Relationships.Remote
import Hasura.RQL.Types.Roles
import Hasura.RQL.Types.SchemaCache
import Hasura.RQL.Types.SchemaCache.Build
import Hasura.RQL.Types.Source
import Hasura.RemoteSchema.Metadata
import Hasura.SQL.Backend
import Hasura.SQL.BackendMap (BackendMap)
import Hasura.SQL.BackendMap qualified as BackendMap
import Hasura.Server.Types
import Hasura.Session
import Network.HTTP.Client.Manager (HasHttpManagerM (..))
import Network.HTTP.Client.Transformable qualified as HTTP
newtype BackendInvalidationKeysWrapper (b :: BackendType) = BackendInvalidationKeysWrapper
{ unBackendInvalidationKeysWrapper :: BackendInvalidationKeys b
}
deriving newtype instance Eq (BackendInvalidationKeys b) => Eq (BackendInvalidationKeysWrapper b)
deriving newtype instance Ord (BackendInvalidationKeys b) => Ord (BackendInvalidationKeysWrapper b)
deriving newtype instance Show (BackendInvalidationKeys b) => Show (BackendInvalidationKeysWrapper b)
deriving newtype instance Semigroup (BackendInvalidationKeys b) => Semigroup (BackendInvalidationKeysWrapper b)
deriving newtype instance Monoid (BackendInvalidationKeys b) => Monoid (BackendInvalidationKeysWrapper b)
instance Inc.Select (BackendInvalidationKeysWrapper b)
-- | 'InvalidationKeys' used to apply requested 'CacheInvalidations'.
data InvalidationKeys = InvalidationKeys
{ _ikMetadata :: Inc.InvalidationKey,
_ikRemoteSchemas :: HashMap RemoteSchemaName Inc.InvalidationKey,
_ikSources :: HashMap SourceName Inc.InvalidationKey,
_ikBackends :: BackendMap BackendInvalidationKeysWrapper
}
deriving (Show, Eq, Generic)
instance Inc.Select InvalidationKeys
$(makeLenses ''InvalidationKeys)
initialInvalidationKeys :: InvalidationKeys
initialInvalidationKeys = InvalidationKeys Inc.initialInvalidationKey mempty mempty mempty
invalidateKeys :: CacheInvalidations -> InvalidationKeys -> InvalidationKeys
invalidateKeys CacheInvalidations {..} InvalidationKeys {..} =
InvalidationKeys
{ _ikMetadata = if ciMetadata then Inc.invalidate _ikMetadata else _ikMetadata,
_ikRemoteSchemas = foldl' (flip invalidate) _ikRemoteSchemas ciRemoteSchemas,
_ikSources = foldl' (flip invalidate) _ikSources ciSources,
_ikBackends = BackendMap.modify @'DataConnector invalidateDataConnectors _ikBackends
}
where
invalidate ::
Hashable a =>
a ->
HashMap a Inc.InvalidationKey ->
HashMap a Inc.InvalidationKey
invalidate = M.alter $ Just . maybe Inc.initialInvalidationKey Inc.invalidate
invalidateDataConnectors :: BackendInvalidationKeysWrapper 'DataConnector -> BackendInvalidationKeysWrapper 'DataConnector
invalidateDataConnectors (BackendInvalidationKeysWrapper invalidationKeys) =
BackendInvalidationKeysWrapper $ foldl' (flip invalidate) invalidationKeys ciDataConnectors
data TableBuildInput b = TableBuildInput
{ _tbiName :: TableName b,
_tbiIsEnum :: Bool,
_tbiConfiguration :: TableConfig b,
_tbiApolloFederationConfig :: Maybe ApolloFederationConfig
}
deriving (Show, Eq, Generic)
instance (Backend b) => NFData (TableBuildInput b)
data NonColumnTableInputs b = NonColumnTableInputs
{ _nctiTable :: TableName b,
_nctiObjectRelationships :: [ObjRelDef b],
_nctiArrayRelationships :: [ArrRelDef b],
_nctiComputedFields :: [ComputedFieldMetadata b],
_nctiRemoteRelationships :: [RemoteRelationship]
}
deriving (Show, Eq, Generic)
data TablePermissionInputs b = TablePermissionInputs
{ _tpiTable :: TableName b,
_tpiInsert :: [InsPermDef b],
_tpiSelect :: [SelPermDef b],
_tpiUpdate :: [UpdPermDef b],
_tpiDelete :: [DelPermDef b]
}
deriving (Generic)
deriving instance (Backend b) => Show (TablePermissionInputs b)
deriving instance (Backend b) => Eq (TablePermissionInputs b)
mkTableInputs ::
TableMetadata b -> (TableBuildInput b, NonColumnTableInputs b, TablePermissionInputs b)
mkTableInputs TableMetadata {..} =
(buildInput, nonColumns, permissions)
where
buildInput = TableBuildInput _tmTable _tmIsEnum _tmConfiguration _tmApolloFederationConfig
nonColumns =
NonColumnTableInputs
_tmTable
(OMap.elems _tmObjectRelationships)
(OMap.elems _tmArrayRelationships)
(OMap.elems _tmComputedFields)
(OMap.elems _tmRemoteRelationships)
permissions =
TablePermissionInputs
_tmTable
(OMap.elems _tmInsertPermissions)
(OMap.elems _tmSelectPermissions)
(OMap.elems _tmUpdatePermissions)
(OMap.elems _tmDeletePermissions)
-- | The direct output of 'buildSchemaCacheRule'. Contains most of the things necessary to build a
-- schema cache, but dependencies and inconsistent metadata objects are collected via a separate
-- 'MonadWriter' side channel.
--
-- See also Note [Avoiding GraphQL schema rebuilds when changing irrelevant Metadata]
data BuildOutputs = BuildOutputs
{ _boSources :: SourceCache,
_boActions :: ActionCache,
-- | We preserve the 'MetadataObject' from the original catalog metadata in the output so we can
-- reuse it later if we need to mark the remote schema inconsistent during GraphQL schema
-- generation (because of field conflicts).
_boRemoteSchemas :: HashMap RemoteSchemaName (RemoteSchemaCtx, MetadataObject),
_boCustomTypes :: AnnotatedCustomTypes,
_boRoles :: HashMap RoleName Role,
_boBackendCache :: BackendCache
}
$(makeLenses ''BuildOutputs)
-- | Parameters required for schema cache build
data CacheBuildParams = CacheBuildParams
{ _cbpManager :: HTTP.Manager,
_cbpPGSourceResolver :: SourceResolver ('Postgres 'Vanilla),
_cbpMSSQLSourceResolver :: SourceResolver 'MSSQL,
_cbpServerConfigCtx :: ServerConfigCtx
}
-- | The monad in which @'RebuildableSchemaCache' is being run
newtype CacheBuild a = CacheBuild (ReaderT CacheBuildParams (ExceptT QErr IO) a)
deriving
( Functor,
Applicative,
Monad,
MonadError QErr,
MonadReader CacheBuildParams,
MonadIO,
MonadBase IO,
MonadBaseControl IO,
MonadUnique
)
instance HasHttpManagerM CacheBuild where
askHttpManager = asks _cbpManager
instance HasServerConfigCtx CacheBuild where
askServerConfigCtx = asks _cbpServerConfigCtx
instance MonadResolveSource CacheBuild where
getPGSourceResolver = asks _cbpPGSourceResolver
getMSSQLSourceResolver = asks _cbpMSSQLSourceResolver
runCacheBuild ::
( MonadIO m,
MonadError QErr m
) =>
CacheBuildParams ->
CacheBuild a ->
m a
runCacheBuild params (CacheBuild m) = do
liftEitherM $ liftIO $ runExceptT (runReaderT m params)
runCacheBuildM ::
( MonadIO m,
MonadError QErr m,
HasHttpManagerM m,
HasServerConfigCtx m,
MonadResolveSource m
) =>
CacheBuild a ->
m a
runCacheBuildM m = do
params <-
CacheBuildParams
<$> askHttpManager
<*> getPGSourceResolver
<*> getMSSQLSourceResolver
<*> askServerConfigCtx
runCacheBuild params m
data RebuildableSchemaCache = RebuildableSchemaCache
{ lastBuiltSchemaCache :: SchemaCache,
_rscInvalidationMap :: InvalidationKeys,
_rscRebuild :: Inc.Rule (ReaderT BuildReason CacheBuild) (Metadata, InvalidationKeys) SchemaCache
}
bindErrorA ::
(ArrowChoice arr, ArrowKleisli m arr, ArrowError e arr, MonadError e m) =>
arr (m a) a
bindErrorA = liftEitherA <<< arrM \m -> (Right <$> m) `catchError` (pure . Left)
{-# INLINE bindErrorA #-}
withRecordDependencies ::
(ArrowWriter (Seq (Either im MetadataDependency)) arr) =>
WriterA (Seq SchemaDependency) arr (e, s) a ->
arr (e, (MetadataObject, (SchemaObjId, s))) a
withRecordDependencies f = proc (e, (metadataObject, (schemaObjectId, s))) -> do
(result, dependencies) <- runWriterA f -< (e, s)
recordDependencies -< (metadataObject, schemaObjectId, toList dependencies)
returnA -< result
{-# INLINEABLE withRecordDependencies #-}
noDuplicates ::
(MonadWriter (Seq (Either InconsistentMetadata md)) m) =>
(a -> MetadataObject) ->
[a] ->
m (Maybe a)
noDuplicates mkMetadataObject = \case
[] -> pure Nothing
[value] -> pure $ Just value
values@(value : _) -> do
let objectId = _moId $ mkMetadataObject value
definitions = map (_moDefinition . mkMetadataObject) values
tell $ Seq.singleton $ Left (DuplicateObjects objectId definitions)
return Nothing
-- | Processes a list of catalog metadata into a map of processed information, marking any duplicate
-- entries inconsistent.
buildInfoMap ::
( ArrowChoice arr,
Inc.ArrowDistribute arr,
ArrowWriter (Seq (Either InconsistentMetadata md)) arr,
Hashable k
) =>
(a -> k) ->
(a -> MetadataObject) ->
(e, a) `arr` Maybe b ->
(e, [a]) `arr` HashMap k b
buildInfoMap extractKey mkMetadataObject buildInfo = proc (e, infos) -> do
let groupedInfos = M.groupOn extractKey infos
infoMapMaybes <-
(|
Inc.keyed
( \_ duplicateInfos -> do
infoMaybe <- interpretWriter -< noDuplicates mkMetadataObject duplicateInfos
case infoMaybe of
Nothing -> returnA -< Nothing
Just info -> buildInfo -< (e, info)
)
|) groupedInfos
returnA -< catMaybes infoMapMaybes
{-# INLINEABLE buildInfoMap #-}
-- | Like 'buildInfo', but includes each processed infos associated 'MetadataObject' in the result.
-- This is useful if the results will be further processed, and the 'MetadataObject' is still needed
-- to mark the object inconsistent.
buildInfoMapPreservingMetadata ::
( ArrowChoice arr,
Inc.ArrowDistribute arr,
ArrowWriter (Seq (Either InconsistentMetadata MetadataDependency)) arr,
Hashable k
) =>
(a -> k) ->
(a -> MetadataObject) ->
(e, a) `arr` Maybe b ->
(e, [a]) `arr` HashMap k (b, MetadataObject)
buildInfoMapPreservingMetadata extractKey mkMetadataObject buildInfo =
buildInfoMap extractKey mkMetadataObject buildInfoPreserving
where
buildInfoPreserving = proc (e, info) -> do
result <- buildInfo -< (e, info)
returnA -< result <&> (,mkMetadataObject info)
{-# INLINEABLE buildInfoMapPreservingMetadata #-}
addTableContext :: (Backend b) => TableName b -> Text -> Text
addTableContext tableName e = "in table " <> tableName <<> ": " <> e