gotosocial/internal/cache/db.go

1658 lines
41 KiB
Go
Raw Normal View History

// GoToSocial
// Copyright (C) GoToSocial Authors admin@gotosocial.org
// SPDX-License-Identifier: AGPL-3.0-or-later
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU Affero General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Affero General Public License for more details.
//
// You should have received a copy of the GNU Affero General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
package cache
import (
"sync/atomic"
"codeberg.org/gruf/go-structr"
"github.com/superseriousbusiness/gotosocial/internal/cache/domain"
"github.com/superseriousbusiness/gotosocial/internal/config"
"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
"github.com/superseriousbusiness/gotosocial/internal/log"
)
type DBCaches struct {
// Account provides access to the gtsmodel Account database cache.
Account StructCache[*gtsmodel.Account]
// AccountNote provides access to the gtsmodel Note database cache.
AccountNote StructCache[*gtsmodel.AccountNote]
// AccountSettings provides access to the gtsmodel AccountSettings database cache.
AccountSettings StructCache[*gtsmodel.AccountSettings]
// AccountStats provides access to the gtsmodel AccountStats database cache.
AccountStats StructCache[*gtsmodel.AccountStats]
// Application provides access to the gtsmodel Application database cache.
Application StructCache[*gtsmodel.Application]
// Block provides access to the gtsmodel Block (account) database cache.
Block StructCache[*gtsmodel.Block]
// BlockIDs provides access to the block IDs database cache.
BlockIDs SliceCache[string]
// BoostOfIDs provides access to the boost of IDs list database cache.
BoostOfIDs SliceCache[string]
// Client provides access to the gtsmodel Client database cache.
Client StructCache[*gtsmodel.Client]
[feature] Conversations API (#3013) * Implement conversations API * Sort and page conversations by last status ID * Appease linter * Fix deleting conversations and statuses * Refactor to make migrations automatic * Lint * Update tests post-merge * Fixes from live-fire testing * Linter caught a format problem * Refactor tests, fix cache * Negative test for non-DMs * Run conversations advanced migration on testrig startup as well as regular server startup * Document (lack of) side effects of API method for deleting a conversation * Make not-found check less nested for readability * Rename PutConversation to UpsertConversation * Use util.Ptr instead of IIFE * Reduce cache used by conversations * Remove unnecessary TableExpr/ColumnExpr * Use struct tags for both unique constraints on Conversation * Make it clear how paging with GetDirectStatusIDsBatch should be used * Let conversation paging skip conversations it can't render * Use Bun NewDropTable * Convert delete raw query to Bun * Convert update raw query to Bun * Convert latestConversationStatusesTempTable raw query partially to Bun * Convert conversationStatusesTempTable raw query partially to Bun * Rename field used to store result of MaxDirectStatusID * Move advanced migrations to their own tiny processor * Catch up util function name with main * Remove json.… wrappers * Remove redundant check * Combine error checks * Replace map with slice of structs * Address processor/type converter comments - Add context info for errors - Extract some common processor code into shared methods - Move conversation eligibility check ahead of populating conversation * Add error context when dropping temp tables
2024-07-23 19:44:31 +00:00
// Conversation provides access to the gtsmodel Conversation database cache.
Conversation StructCache[*gtsmodel.Conversation]
// ConversationLastStatusIDs provides access to the conversation last status IDs database cache.
ConversationLastStatusIDs SliceCache[string]
// DomainAllow provides access to the domain allow database cache.
DomainAllow *domain.Cache
// DomainBlock provides access to the domain block database cache.
DomainBlock *domain.Cache
2024-11-04 16:54:19 +00:00
// DomainPermissionDraft provides access to the domain permission draft database cache.
DomainPermissionDraft StructCache[*gtsmodel.DomainPermissionDraft]
// DomainPermissionSubscription provides access to the domain permission subscription database cache.
DomainPermissionSubscription StructCache[*gtsmodel.DomainPermissionSubscription]
// DomainPermissionExclude provides access to the domain permission exclude database cache.
DomainPermissionExclude *domain.Cache
// Emoji provides access to the gtsmodel Emoji database cache.
Emoji StructCache[*gtsmodel.Emoji]
// EmojiCategory provides access to the gtsmodel EmojiCategory database cache.
EmojiCategory StructCache[*gtsmodel.EmojiCategory]
// Filter provides access to the gtsmodel Filter database cache.
Filter StructCache[*gtsmodel.Filter]
// FilterKeyword provides access to the gtsmodel FilterKeyword database cache.
FilterKeyword StructCache[*gtsmodel.FilterKeyword]
// FilterStatus provides access to the gtsmodel FilterStatus database cache.
FilterStatus StructCache[*gtsmodel.FilterStatus]
// Follow provides access to the gtsmodel Follow database cache.
Follow StructCache[*gtsmodel.Follow]
// FollowIDs provides access to the follower / following IDs database cache.
// THIS CACHE IS KEYED AS THE FOLLOWING {prefix}{accountID} WHERE PREFIX IS:
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
//
// - '>{$accountID}' for following IDs
// e.g. FollowIDs.Load(">" + account.ID, func() {})
// which will load a slice of follows IDs FROM account.
//
// - 'l>{$accountID}' for local following IDs
// e.g. FollowIDs.Load("l>" + account.ID, func() {})
// which will load a slice of LOCAL follows IDs FROM account.
//
// - '<{$accountID}' for follower IDs
// e.g. FollowIDs.Load("<" + account.ID, func() {})
// which will load a slice of follows IDs TARGETTING account.
//
// - 'l<{$accountID}' for local follower IDs
// e.g. FollowIDs.Load("l<" + account.ID, func() {})
// which will load a slice of LOCAL follows IDs TARGETTING account.
//
FollowIDs SliceCache[string]
// FollowRequest provides access to the gtsmodel FollowRequest database cache.
FollowRequest StructCache[*gtsmodel.FollowRequest]
// FollowRequestIDs provides access to the follow requester / requesting IDs database
// cache. THIS CACHE IS KEYED AS THE FOLLOWING {prefix}{accountID} WHERE PREFIX IS:
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
//
// - '>{$accountID}' for follow request IDs
// e.g. FollowRequestIDs.Load(">" + account.ID, func() {})
// which will load a slice of follow request IDs TARGETTING account.
//
// - '<{$accountID}' for follow request IDs
// e.g. FollowRequestIDs.Load("<" + account.ID, func() {})
// which will load a slice of follow request IDs FROM account.
//
FollowRequestIDs SliceCache[string]
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
// FollowingTagIDs provides access to account IDs following / tag IDs followed by
// account db cache. THIS CACHE IS KEYED AS THE FOLLOWING {prefix}{id} WHERE:
//
// - '>{$accountID}' for tag IDs followed by account
// e.g. FollowingTagIDs.Load(">" + account.ID, func() {})
// which will load a slice of tag IDs followed by account.
//
// - '<{$tagIDs}' for account IDs following tag
// e.g. FollowingTagIDs.Load("<" + tag.ID, func() {})
// which will load a slice of account IDs following tag.
//
FollowingTagIDs SliceCache[string]
// Instance provides access to the gtsmodel Instance database cache.
Instance StructCache[*gtsmodel.Instance]
// LocalInstance provides caching for
// simple + common local instance queries.
LocalInstance struct {
Domains atomic.Pointer[int]
Statuses atomic.Pointer[int]
Users atomic.Pointer[int]
}
// InteractionRequest provides access to the gtsmodel InteractionRequest database cache.
InteractionRequest StructCache[*gtsmodel.InteractionRequest]
// InReplyToIDs provides access to the status in reply to IDs list database cache.
InReplyToIDs SliceCache[string]
// List provides access to the gtsmodel List database cache.
List StructCache[*gtsmodel.List]
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
// ListIDs provides access to the list IDs owned by account / list IDs follow
// contained in db cache. THIS CACHE IS KEYED AS FOLLOWING {prefix}{id} WHERE:
//
// - 'a{$accountID}' for list IDs owned by account
// e.g. ListIDs.Load("a" + account.ID, func() {})
// which will load a slice of list IDs owned by account.
//
// - 'f{$followID}' for list IDs follow contained in
// e.g. ListIDs.Load("f" + follow.ID, func() {})
// which will load a slice of list IDs containing follow.
//
ListIDs SliceCache[string]
// ListedIDs provides access to the account IDs in list / follow IDs in
// list db cache. THIS CACHE IS KEYED AS FOLLOWING {prefix}{id} WHERE:
//
// - 'a{listID}' for account IDs in list ID
// e.g. ListedIDs.Load("a" + list.ID, func() {})
// which will load a slice of account IDs in list.
//
// - 'f{listID}' for follow IDs in list ID
// e.g. ListedIDs.Load("f" + list.ID, func() {})
// which will load a slice of follow IDs in list.
//
ListedIDs SliceCache[string]
// Marker provides access to the gtsmodel Marker database cache.
Marker StructCache[*gtsmodel.Marker]
// Media provides access to the gtsmodel Media database cache.
Media StructCache[*gtsmodel.MediaAttachment]
// Mention provides access to the gtsmodel Mention database cache.
Mention StructCache[*gtsmodel.Mention]
// Move provides access to the gtsmodel Move database cache.
Move StructCache[*gtsmodel.Move]
// Notification provides access to the gtsmodel Notification database cache.
Notification StructCache[*gtsmodel.Notification]
// Poll provides access to the gtsmodel Poll database cache.
Poll StructCache[*gtsmodel.Poll]
// PollVote provides access to the gtsmodel PollVote database cache.
PollVote StructCache[*gtsmodel.PollVote]
// PollVoteIDs provides access to the poll vote IDs list database cache.
PollVoteIDs SliceCache[string]
// Report provides access to the gtsmodel Report database cache.
Report StructCache[*gtsmodel.Report]
// SinBinStatus provides access to the gtsmodel SinBinStatus database cache.
SinBinStatus StructCache[*gtsmodel.SinBinStatus]
// Status provides access to the gtsmodel Status database cache.
Status StructCache[*gtsmodel.Status]
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
// StatusBookmark provides access to the gtsmodel StatusBookmark database cache.
StatusBookmark StructCache[*gtsmodel.StatusBookmark]
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
// StatusBookmarkIDs provides access to the status bookmark IDs list database cache.
StatusBookmarkIDs SliceCache[string]
// StatusFave provides access to the gtsmodel StatusFave database cache.
StatusFave StructCache[*gtsmodel.StatusFave]
// StatusFaveIDs provides access to the status fave IDs list database cache.
StatusFaveIDs SliceCache[string]
// Tag provides access to the gtsmodel Tag database cache.
Tag StructCache[*gtsmodel.Tag]
// ThreadMute provides access to the gtsmodel ThreadMute database cache.
ThreadMute StructCache[*gtsmodel.ThreadMute]
// Token provides access to the gtsmodel Token database cache.
Token StructCache[*gtsmodel.Token]
// Tombstone provides access to the gtsmodel Tombstone database cache.
Tombstone StructCache[*gtsmodel.Tombstone]
// User provides access to the gtsmodel User database cache.
User StructCache[*gtsmodel.User]
// UserMute provides access to the gtsmodel UserMute database cache.
UserMute StructCache[*gtsmodel.UserMute]
// UserMuteIDs provides access to the user mute IDs database cache.
UserMuteIDs SliceCache[string]
}
// NOTE:
// all of the below init functions
// are receivers to the main cache
// struct type, not the database cache
// struct type, in order to get access
// to the full suite of caches for
// our invalidate function hooks.
func (c *Caches) initAccount() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofAccount(), // model in-mem size.
config.GetCacheAccountMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(a1 *gtsmodel.Account) *gtsmodel.Account {
a2 := new(gtsmodel.Account)
*a2 = *a1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/account.go.
a2.AvatarMediaAttachment = nil
a2.HeaderMediaAttachment = nil
a2.Emojis = nil
a2.AlsoKnownAs = nil
a2.Move = nil
a2.MovedTo = nil
a2.Settings = nil
a2.Stats = nil
return a2
}
c.DB.Account.Init(structr.CacheConfig[*gtsmodel.Account]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "URL"},
{Fields: "Username,Domain", AllowZero: true},
{Fields: "PublicKeyURI"},
{Fields: "InboxURI"},
{Fields: "OutboxURI"},
{Fields: "FollowersURI"},
{Fields: "FollowingURI"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateAccount,
})
}
func (c *Caches) initAccountNote() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofAccountNote(), // model in-mem size.
config.GetCacheAccountNoteMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(n1 *gtsmodel.AccountNote) *gtsmodel.AccountNote {
n2 := new(gtsmodel.AccountNote)
*n2 = *n1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/relationship_note.go.
n2.Account = nil
n2.TargetAccount = nil
return n2
}
c.DB.AccountNote.Init(structr.CacheConfig[*gtsmodel.AccountNote]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "AccountID,TargetAccountID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initAccountSettings() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofAccountSettings(), // model in-mem size.
config.GetCacheAccountSettingsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.AccountSettings.Init(structr.CacheConfig[*gtsmodel.AccountSettings]{
Indices: []structr.IndexConfig{
{Fields: "AccountID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: func(s1 *gtsmodel.AccountSettings) *gtsmodel.AccountSettings {
s2 := new(gtsmodel.AccountSettings)
*s2 = *s1
return s2
},
})
}
func (c *Caches) initAccountStats() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofAccountStats(), // model in-mem size.
config.GetCacheAccountStatsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.AccountStats.Init(structr.CacheConfig[*gtsmodel.AccountStats]{
Indices: []structr.IndexConfig{
{Fields: "AccountID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: func(s1 *gtsmodel.AccountStats) *gtsmodel.AccountStats {
s2 := new(gtsmodel.AccountStats)
*s2 = *s1
return s2
},
})
}
func (c *Caches) initApplication() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofApplication(), // model in-mem size.
config.GetCacheApplicationMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(a1 *gtsmodel.Application) *gtsmodel.Application {
a2 := new(gtsmodel.Application)
*a2 = *a1
return a2
}
c.DB.Application.Init(structr.CacheConfig[*gtsmodel.Application]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "ClientID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateApplication,
})
}
func (c *Caches) initBlock() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofBlock(), // model in-mem size.
config.GetCacheBlockMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(b1 *gtsmodel.Block) *gtsmodel.Block {
b2 := new(gtsmodel.Block)
*b2 = *b1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/relationship_block.go.
b2.Account = nil
b2.TargetAccount = nil
return b2
}
c.DB.Block.Init(structr.CacheConfig[*gtsmodel.Block]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "AccountID,TargetAccountID"},
{Fields: "AccountID", Multiple: true},
{Fields: "TargetAccountID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateBlock,
})
}
func (c *Caches) initBlockIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheBlockIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.BlockIDs.Init(0, cap)
}
func (c *Caches) initBoostOfIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheBoostOfIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.BoostOfIDs.Init(0, cap)
}
func (c *Caches) initClient() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofClient(), // model in-mem size.
config.GetCacheClientMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(c1 *gtsmodel.Client) *gtsmodel.Client {
c2 := new(gtsmodel.Client)
*c2 = *c1
return c2
}
c.DB.Client.Init(structr.CacheConfig[*gtsmodel.Client]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateClient,
})
}
[feature] Conversations API (#3013) * Implement conversations API * Sort and page conversations by last status ID * Appease linter * Fix deleting conversations and statuses * Refactor to make migrations automatic * Lint * Update tests post-merge * Fixes from live-fire testing * Linter caught a format problem * Refactor tests, fix cache * Negative test for non-DMs * Run conversations advanced migration on testrig startup as well as regular server startup * Document (lack of) side effects of API method for deleting a conversation * Make not-found check less nested for readability * Rename PutConversation to UpsertConversation * Use util.Ptr instead of IIFE * Reduce cache used by conversations * Remove unnecessary TableExpr/ColumnExpr * Use struct tags for both unique constraints on Conversation * Make it clear how paging with GetDirectStatusIDsBatch should be used * Let conversation paging skip conversations it can't render * Use Bun NewDropTable * Convert delete raw query to Bun * Convert update raw query to Bun * Convert latestConversationStatusesTempTable raw query partially to Bun * Convert conversationStatusesTempTable raw query partially to Bun * Rename field used to store result of MaxDirectStatusID * Move advanced migrations to their own tiny processor * Catch up util function name with main * Remove json.… wrappers * Remove redundant check * Combine error checks * Replace map with slice of structs * Address processor/type converter comments - Add context info for errors - Extract some common processor code into shared methods - Move conversation eligibility check ahead of populating conversation * Add error context when dropping temp tables
2024-07-23 19:44:31 +00:00
func (c *Caches) initConversation() {
cap := calculateResultCacheMax(
sizeofConversation(), // model in-mem size.
config.GetCacheConversationMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(c1 *gtsmodel.Conversation) *gtsmodel.Conversation {
c2 := new(gtsmodel.Conversation)
*c2 = *c1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/conversation.go.
c2.Account = nil
c2.OtherAccounts = nil
c2.LastStatus = nil
return c2
}
c.DB.Conversation.Init(structr.CacheConfig[*gtsmodel.Conversation]{
[feature] Conversations API (#3013) * Implement conversations API * Sort and page conversations by last status ID * Appease linter * Fix deleting conversations and statuses * Refactor to make migrations automatic * Lint * Update tests post-merge * Fixes from live-fire testing * Linter caught a format problem * Refactor tests, fix cache * Negative test for non-DMs * Run conversations advanced migration on testrig startup as well as regular server startup * Document (lack of) side effects of API method for deleting a conversation * Make not-found check less nested for readability * Rename PutConversation to UpsertConversation * Use util.Ptr instead of IIFE * Reduce cache used by conversations * Remove unnecessary TableExpr/ColumnExpr * Use struct tags for both unique constraints on Conversation * Make it clear how paging with GetDirectStatusIDsBatch should be used * Let conversation paging skip conversations it can't render * Use Bun NewDropTable * Convert delete raw query to Bun * Convert update raw query to Bun * Convert latestConversationStatusesTempTable raw query partially to Bun * Convert conversationStatusesTempTable raw query partially to Bun * Rename field used to store result of MaxDirectStatusID * Move advanced migrations to their own tiny processor * Catch up util function name with main * Remove json.… wrappers * Remove redundant check * Combine error checks * Replace map with slice of structs * Address processor/type converter comments - Add context info for errors - Extract some common processor code into shared methods - Move conversation eligibility check ahead of populating conversation * Add error context when dropping temp tables
2024-07-23 19:44:31 +00:00
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "ThreadID,AccountID,OtherAccountsKey"},
{Fields: "AccountID,LastStatusID"},
{Fields: "AccountID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateConversation,
})
}
func (c *Caches) initConversationLastStatusIDs() {
cap := calculateSliceCacheMax(
config.GetCacheConversationLastStatusIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.ConversationLastStatusIDs.Init(0, cap)
[feature] Conversations API (#3013) * Implement conversations API * Sort and page conversations by last status ID * Appease linter * Fix deleting conversations and statuses * Refactor to make migrations automatic * Lint * Update tests post-merge * Fixes from live-fire testing * Linter caught a format problem * Refactor tests, fix cache * Negative test for non-DMs * Run conversations advanced migration on testrig startup as well as regular server startup * Document (lack of) side effects of API method for deleting a conversation * Make not-found check less nested for readability * Rename PutConversation to UpsertConversation * Use util.Ptr instead of IIFE * Reduce cache used by conversations * Remove unnecessary TableExpr/ColumnExpr * Use struct tags for both unique constraints on Conversation * Make it clear how paging with GetDirectStatusIDsBatch should be used * Let conversation paging skip conversations it can't render * Use Bun NewDropTable * Convert delete raw query to Bun * Convert update raw query to Bun * Convert latestConversationStatusesTempTable raw query partially to Bun * Convert conversationStatusesTempTable raw query partially to Bun * Rename field used to store result of MaxDirectStatusID * Move advanced migrations to their own tiny processor * Catch up util function name with main * Remove json.… wrappers * Remove redundant check * Combine error checks * Replace map with slice of structs * Address processor/type converter comments - Add context info for errors - Extract some common processor code into shared methods - Move conversation eligibility check ahead of populating conversation * Add error context when dropping temp tables
2024-07-23 19:44:31 +00:00
}
func (c *Caches) initDomainAllow() {
c.DB.DomainAllow = new(domain.Cache)
}
func (c *Caches) initDomainBlock() {
c.DB.DomainBlock = new(domain.Cache)
}
2024-11-04 16:54:19 +00:00
func (c *Caches) initDomainPermissionDraft() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofDomainPermissionDraft(), // model in-mem size.
config.GetCacheDomainPermissionDraftMemRation(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(d1 *gtsmodel.DomainPermissionDraft) *gtsmodel.DomainPermissionDraft {
d2 := new(gtsmodel.DomainPermissionDraft)
*d2 = *d1
// Don't include ptr fields that
// will be populated separately.
d2.CreatedByAccount = nil
return d2
}
c.DB.DomainPermissionDraft.Init(structr.CacheConfig[*gtsmodel.DomainPermissionDraft]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "Domain", Multiple: true},
{Fields: "SubscriptionID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initDomainPermissionSubscription() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofDomainPermissionSubscription(), // model in-mem size.
config.GetCacheDomainPermissionSubscriptionMemRation(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(d1 *gtsmodel.DomainPermissionSubscription) *gtsmodel.DomainPermissionSubscription {
d2 := new(gtsmodel.DomainPermissionSubscription)
*d2 = *d1
// Don't include ptr fields that
// will be populated separately.
d2.CreatedByAccount = nil
return d2
}
c.DB.DomainPermissionSubscription.Init(structr.CacheConfig[*gtsmodel.DomainPermissionSubscription]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initDomainPermissionExclude() {
c.DB.DomainPermissionExclude = new(domain.Cache)
}
func (c *Caches) initEmoji() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofEmoji(), // model in-mem size.
config.GetCacheEmojiMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(e1 *gtsmodel.Emoji) *gtsmodel.Emoji {
e2 := new(gtsmodel.Emoji)
*e2 = *e1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/emoji.go.
e2.Category = nil
return e2
}
c.DB.Emoji.Init(structr.CacheConfig[*gtsmodel.Emoji]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "Shortcode,Domain", AllowZero: true},
{Fields: "ImageStaticURL"},
{Fields: "CategoryID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initEmojiCategory() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofEmojiCategory(), // model in-mem size.
config.GetCacheEmojiCategoryMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(c1 *gtsmodel.EmojiCategory) *gtsmodel.EmojiCategory {
c2 := new(gtsmodel.EmojiCategory)
*c2 = *c1
return c2
}
c.DB.EmojiCategory.Init(structr.CacheConfig[*gtsmodel.EmojiCategory]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "Name"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateEmojiCategory,
})
}
func (c *Caches) initFilter() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofFilter(), // model in-mem size.
config.GetCacheFilterMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(filter1 *gtsmodel.Filter) *gtsmodel.Filter {
filter2 := new(gtsmodel.Filter)
*filter2 = *filter1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/filter.go.
filter2.Keywords = nil
filter2.Statuses = nil
return filter2
}
c.DB.Filter.Init(structr.CacheConfig[*gtsmodel.Filter]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "AccountID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initFilterKeyword() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofFilterKeyword(), // model in-mem size.
config.GetCacheFilterKeywordMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(filterKeyword1 *gtsmodel.FilterKeyword) *gtsmodel.FilterKeyword {
filterKeyword2 := new(gtsmodel.FilterKeyword)
*filterKeyword2 = *filterKeyword1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/filter.go.
filterKeyword2.Filter = nil
// We specifically DO NOT unset
// the regexp field here, as any
// regexp.Regexp instance is safe
// for concurrent access.
return filterKeyword2
}
c.DB.FilterKeyword.Init(structr.CacheConfig[*gtsmodel.FilterKeyword]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "AccountID", Multiple: true},
{Fields: "FilterID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initFilterStatus() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofFilterStatus(), // model in-mem size.
config.GetCacheFilterStatusMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(filterStatus1 *gtsmodel.FilterStatus) *gtsmodel.FilterStatus {
filterStatus2 := new(gtsmodel.FilterStatus)
*filterStatus2 = *filterStatus1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/filter.go.
filterStatus2.Filter = nil
return filterStatus2
}
c.DB.FilterStatus.Init(structr.CacheConfig[*gtsmodel.FilterStatus]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "AccountID", Multiple: true},
{Fields: "FilterID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initFollow() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofFollow(), // model in-mem size.
config.GetCacheFollowMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(f1 *gtsmodel.Follow) *gtsmodel.Follow {
f2 := new(gtsmodel.Follow)
*f2 = *f1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/relationship_follow.go.
f2.Account = nil
f2.TargetAccount = nil
return f2
}
c.DB.Follow.Init(structr.CacheConfig[*gtsmodel.Follow]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "AccountID,TargetAccountID"},
{Fields: "AccountID", Multiple: true},
{Fields: "TargetAccountID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateFollow,
})
}
func (c *Caches) initFollowIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheFollowIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.FollowIDs.Init(0, cap)
}
func (c *Caches) initFollowRequest() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofFollowRequest(), // model in-mem size.
config.GetCacheFollowRequestMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(f1 *gtsmodel.FollowRequest) *gtsmodel.FollowRequest {
f2 := new(gtsmodel.FollowRequest)
*f2 = *f1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/relationship_follow_req.go.
f2.Account = nil
f2.TargetAccount = nil
return f2
}
c.DB.FollowRequest.Init(structr.CacheConfig[*gtsmodel.FollowRequest]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "AccountID,TargetAccountID"},
{Fields: "AccountID", Multiple: true},
{Fields: "TargetAccountID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateFollowRequest,
})
}
func (c *Caches) initFollowRequestIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheFollowRequestIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.FollowRequestIDs.Init(0, cap)
}
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
func (c *Caches) initFollowingTagIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheFollowingTagIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.FollowingTagIDs.Init(0, cap)
}
func (c *Caches) initInReplyToIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheInReplyToIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.InReplyToIDs.Init(0, cap)
}
func (c *Caches) initInstance() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofInstance(), // model in-mem size.
config.GetCacheInstanceMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(i1 *gtsmodel.Instance) *gtsmodel.Instance {
i2 := new(gtsmodel.Instance)
*i2 = *i1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/instance.go.
i2.DomainBlock = nil
i2.ContactAccount = nil
return i2
}
c.DB.Instance.Init(structr.CacheConfig[*gtsmodel.Instance]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "Domain"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateInstance,
})
}
func (c *Caches) initInteractionRequest() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofInteractionRequest(),
config.GetCacheInteractionRequestMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(i1 *gtsmodel.InteractionRequest) *gtsmodel.InteractionRequest {
i2 := new(gtsmodel.InteractionRequest)
*i2 = *i1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/interaction.go.
i2.Status = nil
i2.TargetAccount = nil
i2.InteractingAccount = nil
i2.Like = nil
i2.Reply = nil
i2.Announce = nil
return i2
}
c.DB.InteractionRequest.Init(structr.CacheConfig[*gtsmodel.InteractionRequest]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "InteractionURI"},
{Fields: "URI"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initList() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofList(), // model in-mem size.
config.GetCacheListMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(l1 *gtsmodel.List) *gtsmodel.List {
l2 := new(gtsmodel.List)
*l2 = *l1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/list.go.
l2.Account = nil
return l2
}
c.DB.List.Init(structr.CacheConfig[*gtsmodel.List]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateList,
})
}
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
func (c *Caches) initListIDs() {
// Calculate maximum cache size.
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
cap := calculateSliceCacheMax(
config.GetCacheListIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
c.DB.ListIDs.Init(0, cap)
}
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
func (c *Caches) initListedIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheListedIDsMemRatio(),
)
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
log.Infof(nil, "cache size = %d", cap)
[performance] cache more database calls, reduce required database calls overall (#3290) * improvements to caching for lists and relationship to accounts / follows * fix nil panic in AddToList() * ensure list related caches are correctly invalidated * ensure returned ID lists are ordered correctly * bump go-structr to v0.8.9 (returns early if zero uncached keys to be loaded) * remove zero checks in uncached key load functions (go-structr now handles this) * fix issues after rebase on upstream/main * update the expected return order of CSV exports (since list entries are now down by entry creation date) * rename some funcs, allow deleting list entries for multiple follow IDs at a time, fix up more tests * use returning statements on delete to get cache invalidation info * fixes to recent database delete changes * fix broken list entries delete sql * remove unused db function * update remainder of delete functions to behave in similar way, some other small tweaks * fix delete user sql, allow returning on err no entries * uncomment + fix list database tests * update remaining list tests * update envparsing test * add comments to each specific key being invalidated * add more cache invalidation explanatory comments * whoops; actually delete poll votes from database in the DeletePollByID() func * remove added but-commented-out field * improved comment regarding paging being disabled * make cache invalidation comments match what's actually happening * fix up delete query comments to match what is happening * rename function to read a bit better * don't use ErrNoEntries on delete when not needed (it's only needed for a RETURNING call) * update function name in test * move list exclusivity check to AFTER eligibility check. use log.Panic() instead of panic() * use the poll_id column in poll_votes for selecting votes in poll ID * fix function name
2024-09-16 16:46:09 +00:00
c.DB.ListedIDs.Init(0, cap)
}
func (c *Caches) initMarker() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofMarker(), // model in-mem size.
config.GetCacheMarkerMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(m1 *gtsmodel.Marker) *gtsmodel.Marker {
m2 := new(gtsmodel.Marker)
*m2 = *m1
return m2
}
c.DB.Marker.Init(structr.CacheConfig[*gtsmodel.Marker]{
Indices: []structr.IndexConfig{
{Fields: "AccountID,Name"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initMedia() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofMedia(), // model in-mem size.
config.GetCacheMediaMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(m1 *gtsmodel.MediaAttachment) *gtsmodel.MediaAttachment {
m2 := new(gtsmodel.MediaAttachment)
*m2 = *m1
return m2
}
c.DB.Media.Init(structr.CacheConfig[*gtsmodel.MediaAttachment]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateMedia,
})
}
func (c *Caches) initMention() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofMention(), // model in-mem size.
config.GetCacheMentionMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(m1 *gtsmodel.Mention) *gtsmodel.Mention {
m2 := new(gtsmodel.Mention)
*m2 = *m1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/mention.go.
m2.Status = nil
m2.OriginAccount = nil
m2.TargetAccount = nil
return m2
}
c.DB.Mention.Init(structr.CacheConfig[*gtsmodel.Mention]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initMove() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofMove(), // model in-mem size.
config.GetCacheMoveMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.Move.Init(structr.CacheConfig[*gtsmodel.Move]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "OriginURI,TargetURI"},
{Fields: "OriginURI", Multiple: true},
{Fields: "TargetURI", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: func(m1 *gtsmodel.Move) *gtsmodel.Move {
m2 := new(gtsmodel.Move)
*m2 = *m1
return m2
},
})
}
func (c *Caches) initNotification() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofNotification(), // model in-mem size.
config.GetCacheNotificationMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(n1 *gtsmodel.Notification) *gtsmodel.Notification {
n2 := new(gtsmodel.Notification)
*n2 = *n1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/notification.go.
n2.Status = nil
n2.OriginAccount = nil
n2.TargetAccount = nil
return n2
}
c.DB.Notification.Init(structr.CacheConfig[*gtsmodel.Notification]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "NotificationType,TargetAccountID,OriginAccountID,StatusID", AllowZero: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initPoll() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofPoll(), // model in-mem size.
config.GetCachePollMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(p1 *gtsmodel.Poll) *gtsmodel.Poll {
p2 := new(gtsmodel.Poll)
*p2 = *p1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/poll.go.
p2.Status = nil
// Don't include ephemeral fields
// which are only expected to be
// set on ONE poll instance.
p2.Closing = false
return p2
}
c.DB.Poll.Init(structr.CacheConfig[*gtsmodel.Poll]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "StatusID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidatePoll,
})
}
func (c *Caches) initPollVote() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofPollVote(), // model in-mem size.
config.GetCachePollVoteMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(v1 *gtsmodel.PollVote) *gtsmodel.PollVote {
v2 := new(gtsmodel.PollVote)
*v2 = *v1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/poll.go.
v2.Account = nil
v2.Poll = nil
return v2
}
c.DB.PollVote.Init(structr.CacheConfig[*gtsmodel.PollVote]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "PollID", Multiple: true},
{Fields: "PollID,AccountID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidatePollVote,
})
}
func (c *Caches) initPollVoteIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCachePollVoteIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.PollVoteIDs.Init(0, cap)
}
func (c *Caches) initReport() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofReport(), // model in-mem size.
config.GetCacheReportMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(r1 *gtsmodel.Report) *gtsmodel.Report {
r2 := new(gtsmodel.Report)
*r2 = *r1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/report.go.
r2.Account = nil
r2.TargetAccount = nil
r2.Statuses = nil
r2.Rules = nil
r2.ActionTakenByAccount = nil
return r2
}
c.DB.Report.Init(structr.CacheConfig[*gtsmodel.Report]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initSinBinStatus() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofSinBinStatus(), // model in-mem size.
config.GetCacheSinBinStatusMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(s1 *gtsmodel.SinBinStatus) *gtsmodel.SinBinStatus {
s2 := new(gtsmodel.SinBinStatus)
*s2 = *s1
return s2
}
c.DB.SinBinStatus.Init(structr.CacheConfig[*gtsmodel.SinBinStatus]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initStatus() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofStatus(), // model in-mem size.
config.GetCacheStatusMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(s1 *gtsmodel.Status) *gtsmodel.Status {
s2 := new(gtsmodel.Status)
*s2 = *s1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/status.go.
s2.Account = nil
s2.InReplyTo = nil
s2.InReplyToAccount = nil
s2.BoostOf = nil
s2.BoostOfAccount = nil
s2.Poll = nil
s2.Attachments = nil
s2.Tags = nil
s2.Mentions = nil
s2.Emojis = nil
s2.CreatedWithApplication = nil
return s2
}
c.DB.Status.Init(structr.CacheConfig[*gtsmodel.Status]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "URL"},
{Fields: "PollID"},
{Fields: "BoostOfID,AccountID"},
{Fields: "ThreadID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateStatus,
})
}
func (c *Caches) initStatusBookmark() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofStatusBookmark(), // model in-mem size.
config.GetCacheStatusBookmarkMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(s1 *gtsmodel.StatusBookmark) *gtsmodel.StatusBookmark {
s2 := new(gtsmodel.StatusBookmark)
*s2 = *s1
// Don't include ptr fields that
// will be populated separately.
s2.Account = nil
s2.TargetAccount = nil
s2.Status = nil
return s2
}
c.DB.StatusBookmark.Init(structr.CacheConfig[*gtsmodel.StatusBookmark]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "AccountID,StatusID"},
{Fields: "AccountID", Multiple: true},
{Fields: "TargetAccountID", Multiple: true},
{Fields: "StatusID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateStatusBookmark,
})
}
func (c *Caches) initStatusBookmarkIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheStatusBookmarkIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.StatusBookmarkIDs.Init(0, cap)
}
func (c *Caches) initStatusFave() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofStatusFave(), // model in-mem size.
config.GetCacheStatusFaveMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(f1 *gtsmodel.StatusFave) *gtsmodel.StatusFave {
f2 := new(gtsmodel.StatusFave)
*f2 = *f1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/statusfave.go.
f2.Account = nil
f2.TargetAccount = nil
f2.Status = nil
return f2
}
c.DB.StatusFave.Init(structr.CacheConfig[*gtsmodel.StatusFave]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
{Fields: "AccountID,StatusID"},
{Fields: "StatusID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateStatusFave,
})
}
func (c *Caches) initStatusFaveIDs() {
// Calculate maximum cache size.
cap := calculateSliceCacheMax(
config.GetCacheStatusFaveIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.StatusFaveIDs.Init(0, cap)
}
func (c *Caches) initTag() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofTag(), // model in-mem size.
config.GetCacheTagMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(m1 *gtsmodel.Tag) *gtsmodel.Tag {
m2 := new(gtsmodel.Tag)
*m2 = *m1
return m2
}
c.DB.Tag.Init(structr.CacheConfig[*gtsmodel.Tag]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "Name"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initThreadMute() {
cap := calculateResultCacheMax(
sizeofThreadMute(), // model in-mem size.
config.GetCacheThreadMuteMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(t1 *gtsmodel.ThreadMute) *gtsmodel.ThreadMute {
t2 := new(gtsmodel.ThreadMute)
*t2 = *t1
return t2
}
c.DB.ThreadMute.Init(structr.CacheConfig[*gtsmodel.ThreadMute]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "ThreadID", Multiple: true},
{Fields: "AccountID", Multiple: true},
{Fields: "ThreadID,AccountID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initToken() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofToken(), // model in-mem size.
config.GetCacheTokenMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(t1 *gtsmodel.Token) *gtsmodel.Token {
t2 := new(gtsmodel.Token)
*t2 = *t1
return t2
}
c.DB.Token.Init(structr.CacheConfig[*gtsmodel.Token]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "Code"},
{Fields: "Access"},
{Fields: "Refresh"},
{Fields: "ClientID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initTombstone() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofTombstone(), // model in-mem size.
config.GetCacheTombstoneMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(t1 *gtsmodel.Tombstone) *gtsmodel.Tombstone {
t2 := new(gtsmodel.Tombstone)
*t2 = *t1
return t2
}
c.DB.Tombstone.Init(structr.CacheConfig[*gtsmodel.Tombstone]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "URI"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
})
}
func (c *Caches) initUser() {
// Calculate maximum cache size.
cap := calculateResultCacheMax(
sizeofUser(), // model in-mem size.
config.GetCacheUserMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(u1 *gtsmodel.User) *gtsmodel.User {
u2 := new(gtsmodel.User)
*u2 = *u1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/user.go.
u2.Account = nil
return u2
}
c.DB.User.Init(structr.CacheConfig[*gtsmodel.User]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "AccountID"},
{Fields: "Email"},
{Fields: "ConfirmationToken"},
{Fields: "ExternalID"},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateUser,
})
}
func (c *Caches) initUserMute() {
cap := calculateResultCacheMax(
sizeofUserMute(), // model in-mem size.
config.GetCacheUserMuteMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
copyF := func(u1 *gtsmodel.UserMute) *gtsmodel.UserMute {
u2 := new(gtsmodel.UserMute)
*u2 = *u1
// Don't include ptr fields that
// will be populated separately.
// See internal/db/bundb/relationship_mute.go.
u2.Account = nil
u2.TargetAccount = nil
return u2
}
c.DB.UserMute.Init(structr.CacheConfig[*gtsmodel.UserMute]{
Indices: []structr.IndexConfig{
{Fields: "ID"},
{Fields: "AccountID,TargetAccountID"},
{Fields: "AccountID", Multiple: true},
{Fields: "TargetAccountID", Multiple: true},
},
MaxSize: cap,
IgnoreErr: ignoreErrors,
Copy: copyF,
Invalidate: c.OnInvalidateUserMute,
})
}
func (c *Caches) initUserMuteIDs() {
cap := calculateSliceCacheMax(
config.GetCacheUserMuteIDsMemRatio(),
)
log.Infof(nil, "cache size = %d", cap)
c.DB.UserMuteIDs.Init(0, cap)
}