forgejo/models/db/engine.go

331 lines
9.4 KiB
Go
Raw Normal View History

2014-02-12 17:49:46 +00:00
// Copyright 2014 The Gogs Authors. All rights reserved.
// Copyright 2018 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
2014-02-12 17:49:46 +00:00
package db
2014-02-18 22:48:02 +00:00
import (
"context"
2014-10-19 05:35:24 +00:00
"database/sql"
2014-02-18 22:48:02 +00:00
"fmt"
"io"
"reflect"
"strings"
[GITEA] Add slow SQL query warning - Databases are one of the most important parts of Forgejo, every interaction with Forgejo uses the database in one way or another. Therefore, it is important to maintain the database and recognize when Forgejo is not doing well with the database. Forgejo already has the option to log *every* SQL query along with its execution time, but monitoring becomes impractical for larger instances and takes up unnecessary storage in the logs. - Add a QoL enhancement that allows instance administrators to specify a threshold value beyond which query execution time is logged as a warning in the xorm logger. The default value is a conservative five seconds to avoid this becoming a source of spam in the logs. - The use case for this patch is that with an instance the size of Codeberg, monitoring SQL logs is not very fruitful and most of them are uninteresting. Recently, in the context of persistent deadlock issues (https://codeberg.org/forgejo/forgejo/issues/220), I have noticed that certain queries hold locks on tables like comment and issue for several seconds. This patch helps to identify which queries these are and when they happen. - Added unit test. (cherry picked from commit 24bbe7886fb4cb9a38c8dab8c44f4c9cbfa25481) (cherry picked from commit 6e29145b3c1455498531593d38e6a914941a12cb) (cherry picked from commit 63731e30712872bd2395eb3cf36d9996e5793645) (cherry picked from commit 3ce1a097369c132654de70df707b867e47bd1c40) (cherry picked from commit a64426907de788cc0937a7a2b16af4d2f26f7fe6)
2023-08-18 02:39:23 +00:00
"time"
2014-02-18 22:48:02 +00:00
[GITEA] Add slow SQL query warning - Databases are one of the most important parts of Forgejo, every interaction with Forgejo uses the database in one way or another. Therefore, it is important to maintain the database and recognize when Forgejo is not doing well with the database. Forgejo already has the option to log *every* SQL query along with its execution time, but monitoring becomes impractical for larger instances and takes up unnecessary storage in the logs. - Add a QoL enhancement that allows instance administrators to specify a threshold value beyond which query execution time is logged as a warning in the xorm logger. The default value is a conservative five seconds to avoid this becoming a source of spam in the logs. - The use case for this patch is that with an instance the size of Codeberg, monitoring SQL logs is not very fruitful and most of them are uninteresting. Recently, in the context of persistent deadlock issues (https://codeberg.org/forgejo/forgejo/issues/220), I have noticed that certain queries hold locks on tables like comment and issue for several seconds. This patch helps to identify which queries these are and when they happen. - Added unit test. (cherry picked from commit 24bbe7886fb4cb9a38c8dab8c44f4c9cbfa25481) (cherry picked from commit 6e29145b3c1455498531593d38e6a914941a12cb) (cherry picked from commit 63731e30712872bd2395eb3cf36d9996e5793645) (cherry picked from commit 3ce1a097369c132654de70df707b867e47bd1c40) (cherry picked from commit a64426907de788cc0937a7a2b16af4d2f26f7fe6)
2023-08-18 02:39:23 +00:00
"code.gitea.io/gitea/modules/log"
"code.gitea.io/gitea/modules/setting"
2019-10-17 09:26:49 +00:00
"xorm.io/xorm"
[GITEA] Add slow SQL query warning - Databases are one of the most important parts of Forgejo, every interaction with Forgejo uses the database in one way or another. Therefore, it is important to maintain the database and recognize when Forgejo is not doing well with the database. Forgejo already has the option to log *every* SQL query along with its execution time, but monitoring becomes impractical for larger instances and takes up unnecessary storage in the logs. - Add a QoL enhancement that allows instance administrators to specify a threshold value beyond which query execution time is logged as a warning in the xorm logger. The default value is a conservative five seconds to avoid this becoming a source of spam in the logs. - The use case for this patch is that with an instance the size of Codeberg, monitoring SQL logs is not very fruitful and most of them are uninteresting. Recently, in the context of persistent deadlock issues (https://codeberg.org/forgejo/forgejo/issues/220), I have noticed that certain queries hold locks on tables like comment and issue for several seconds. This patch helps to identify which queries these are and when they happen. - Added unit test. (cherry picked from commit 24bbe7886fb4cb9a38c8dab8c44f4c9cbfa25481) (cherry picked from commit 6e29145b3c1455498531593d38e6a914941a12cb) (cherry picked from commit 63731e30712872bd2395eb3cf36d9996e5793645) (cherry picked from commit 3ce1a097369c132654de70df707b867e47bd1c40) (cherry picked from commit a64426907de788cc0937a7a2b16af4d2f26f7fe6)
2023-08-18 02:39:23 +00:00
"xorm.io/xorm/contexts"
"xorm.io/xorm/names"
"xorm.io/xorm/schemas"
2016-11-26 00:20:18 +00:00
_ "github.com/denisenkom/go-mssqldb" // Needed for the MSSQL driver
_ "github.com/go-sql-driver/mysql" // Needed for the MySQL driver
_ "github.com/lib/pq" // Needed for the Postgresql driver
2014-02-18 22:48:02 +00:00
)
var (
x *xorm.Engine
tables []any
initFuncs []func() error
// HasEngine specifies if we have a xorm.Engine
HasEngine bool
)
2014-10-19 05:35:24 +00:00
// Engine represents a xorm engine or session.
type Engine interface {
Table(tableNameOrBean any) *xorm.Session
Count(...any) (int64, error)
Decr(column string, arg ...any) *xorm.Session
Delete(...any) (int64, error)
Truncate(...any) (int64, error)
Exec(...any) (sql.Result, error)
Find(any, ...any) error
Get(beans ...any) (bool, error)
ID(any) *xorm.Session
In(string, ...any) *xorm.Session
Incr(column string, arg ...any) *xorm.Session
Insert(...any) (int64, error)
Iterate(any, xorm.IterFunc) error
IsTableExist(any) (bool, error)
Join(joinOperator string, tablename, condition any, args ...any) *xorm.Session
SQL(any, ...any) *xorm.Session
Where(any, ...any) *xorm.Session
2019-06-12 19:41:28 +00:00
Asc(colNames ...string) *xorm.Session
Desc(colNames ...string) *xorm.Session
API add/generalize pagination (#9452) * paginate results * fixed deadlock * prevented breaking change * updated swagger * go fmt * fixed find topic * go mod tidy * go mod vendor with go1.13.5 * fixed repo find topics * fixed unit test * added Limit method to Engine struct; use engine variable when provided; fixed gitignore * use ItemsPerPage for default pagesize; fix GetWatchers, getOrgUsersByOrgID and GetStargazers; fix GetAllCommits headers; reverted some changed behaviors * set Page value on Home route * improved memory allocations * fixed response headers * removed logfiles * fixed import order * import order * improved swagger * added function to get models.ListOptions from context * removed pagesize diff on unit test * fixed imports * removed unnecessary struct field * fixed go fmt * scoped PR * code improvements * code improvements * go mod tidy * fixed import order * fixed commit statuses session * fixed files headers * fixed headers; added pagination for notifications * go mod tidy * go fmt * removed Private from user search options; added setting.UI.IssuePagingNum as default valeu on repo's issues list * Apply suggestions from code review Co-Authored-By: 6543 <6543@obermui.de> Co-Authored-By: zeripath <art27@cantab.net> * fixed build error * CI.restart() * fixed merge conflicts resolve * fixed conflicts resolve * improved FindTrackedTimesOptions.ToOptions() method * added backwards compatibility on ListReleases request; fixed issue tracked time ToSession * fixed build error; fixed swagger template * fixed swagger template * fixed ListReleases backwards compatibility * added page to user search route Co-authored-by: techknowlogick <matti@mdranta.net> Co-authored-by: 6543 <6543@obermui.de> Co-authored-by: zeripath <art27@cantab.net>
2020-01-24 19:00:29 +00:00
Limit(limit int, start ...int) *xorm.Session
NoAutoTime() *xorm.Session
SumInt(bean any, columnName string) (res int64, err error)
Sync(...any) error
Select(string) *xorm.Session
NotIn(string, ...any) *xorm.Session
OrderBy(any, ...any) *xorm.Session
Exist(...any) (bool, error)
Distinct(...string) *xorm.Session
Query(...any) ([]map[string][]byte, error)
Cols(...string) *xorm.Session
Context(ctx context.Context) *xorm.Session
Ping() error
2014-10-19 05:35:24 +00:00
}
// TableInfo returns table's information via an object
func TableInfo(v any) (*schemas.Table, error) {
return x.TableInfo(v)
}
// DumpTables dump tables information
func DumpTables(tables []*schemas.Table, w io.Writer, tp ...schemas.DBType) error {
return x.DumpTables(tables, w, tp...)
}
2016-11-26 00:20:18 +00:00
// RegisterModel registers model, if initfunc provided, it will be invoked after data model sync
func RegisterModel(bean any, initFunc ...func() error) {
tables = append(tables, bean)
if len(initFuncs) > 0 && initFunc[0] != nil {
initFuncs = append(initFuncs, initFunc[0])
}
}
2014-03-21 05:48:10 +00:00
2014-04-05 14:46:32 +00:00
func init() {
2016-11-26 00:20:18 +00:00
gonicNames := []string{"SSL", "UID"}
2015-08-27 15:06:14 +00:00
for _, name := range gonicNames {
names.LintGonicMapper[name] = true
2015-08-27 15:06:14 +00:00
}
2014-04-05 14:46:32 +00:00
}
// newXORMEngine returns a new XORM engine from the configuration
func newXORMEngine() (*xorm.Engine, error) {
connStr, err := setting.DBConnStr()
if err != nil {
return nil, err
2014-03-30 14:47:08 +00:00
}
var engine *xorm.Engine
if setting.Database.Type.IsPostgreSQL() && len(setting.Database.Schema) > 0 {
// OK whilst we sort out our schema issues - create a schema aware postgres
registerPostgresSchemaDriver()
engine, err = xorm.NewEngine("postgresschema", connStr)
} else {
engine, err = xorm.NewEngine(setting.Database.Type.String(), connStr)
}
if err != nil {
return nil, err
}
if setting.Database.Type == "mysql" {
engine.Dialect().SetParams(map[string]string{"rowFormat": "DYNAMIC"})
} else if setting.Database.Type == "mssql" {
engine.Dialect().SetParams(map[string]string{"DEFAULT_VARCHAR": "nvarchar"})
}
engine.SetSchema(setting.Database.Schema)
return engine, nil
}
// SyncAllTables sync the schemas of all tables, is required by unit test code
func SyncAllTables() error {
Fix wrong warn messages in migration steps (#25475) The recent change on xorm for `Sync` is it will not warn when database have columns which is not listed on struct. So we just need this warn logs when `Sync` the whole database but not in the migrations Sync. This PR will remove almost unnecessary warning logs on migrations. Now below logs in CI will disappear. ```log 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column creator_id but struct has not related field 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column is_closed but struct has not related field 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column board_type but struct has not related field 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column type but struct has not related field 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column closed_date_unix but struct has not related field 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column created_unix but struct has not related field 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column updated_unix but struct has not related field 2023/06/23 17:51:32 models/db/engine.go:191:InitEngineWithMigration() [W] Table gtestschema.project has column card_type but struct has not related field ```
2023-06-24 08:20:08 +00:00
_, err := x.StoreEngine("InnoDB").SyncWithOptions(xorm.SyncOptions{
WarnIfDatabaseColumnMissed: true,
}, tables...)
return err
}
// InitEngine initializes the xorm.Engine and sets it as db.DefaultContext
func InitEngine(ctx context.Context) error {
xormEngine, err := newXORMEngine()
2014-02-18 22:48:02 +00:00
if err != nil {
return fmt.Errorf("failed to connect to database: %w", err)
2014-02-18 22:48:02 +00:00
}
xormEngine.SetMapper(names.GonicMapper{})
2014-12-07 01:22:48 +00:00
// WARNING: for serv command, MUST remove the output to os.stdout,
2014-03-20 20:04:56 +00:00
// so use log file to instead print to stdout.
xormEngine.SetLogger(NewXORMLogger(setting.Database.LogSQL))
xormEngine.ShowSQL(setting.Database.LogSQL)
xormEngine.SetMaxOpenConns(setting.Database.MaxOpenConns)
xormEngine.SetMaxIdleConns(setting.Database.MaxIdleConns)
xormEngine.SetConnMaxLifetime(setting.Database.ConnMaxLifetime)
xormEngine.SetDefaultContext(ctx)
[GITEA] Add slow SQL query warning - Databases are one of the most important parts of Forgejo, every interaction with Forgejo uses the database in one way or another. Therefore, it is important to maintain the database and recognize when Forgejo is not doing well with the database. Forgejo already has the option to log *every* SQL query along with its execution time, but monitoring becomes impractical for larger instances and takes up unnecessary storage in the logs. - Add a QoL enhancement that allows instance administrators to specify a threshold value beyond which query execution time is logged as a warning in the xorm logger. The default value is a conservative five seconds to avoid this becoming a source of spam in the logs. - The use case for this patch is that with an instance the size of Codeberg, monitoring SQL logs is not very fruitful and most of them are uninteresting. Recently, in the context of persistent deadlock issues (https://codeberg.org/forgejo/forgejo/issues/220), I have noticed that certain queries hold locks on tables like comment and issue for several seconds. This patch helps to identify which queries these are and when they happen. - Added unit test. (cherry picked from commit 24bbe7886fb4cb9a38c8dab8c44f4c9cbfa25481) (cherry picked from commit 6e29145b3c1455498531593d38e6a914941a12cb) (cherry picked from commit 63731e30712872bd2395eb3cf36d9996e5793645) (cherry picked from commit 3ce1a097369c132654de70df707b867e47bd1c40) (cherry picked from commit a64426907de788cc0937a7a2b16af4d2f26f7fe6)
2023-08-18 02:39:23 +00:00
if setting.Database.SlowQueryTreshold > 0 {
xormEngine.AddHook(&SlowQueryHook{
Treshold: setting.Database.SlowQueryTreshold,
Logger: log.GetLogger("xorm"),
})
}
SetDefaultEngine(ctx, xormEngine)
return nil
}
// SetDefaultEngine sets the default engine for db
func SetDefaultEngine(ctx context.Context, eng *xorm.Engine) {
x = eng
DefaultContext = &Context{
Context: ctx,
e: x,
}
2014-02-18 22:48:02 +00:00
}
// UnsetDefaultEngine closes and unsets the default engine
// We hope the SetDefaultEngine and UnsetDefaultEngine can be paired, but it's impossible now,
// there are many calls to InitEngine -> SetDefaultEngine directly to overwrite the `x` and DefaultContext without close
// Global database engine related functions are all racy and there is no graceful close right now.
func UnsetDefaultEngine() {
if x != nil {
_ = x.Close()
x = nil
}
DefaultContext = nil
}
// InitEngineWithMigration initializes a new xorm.Engine and sets it as the db.DefaultContext
// This function must never call .Sync() if the provided migration function fails.
// When called from the "doctor" command, the migration function is a version check
// that prevents the doctor from fixing anything in the database if the migration level
// is different from the expected value.
func InitEngineWithMigration(ctx context.Context, migrateFunc func(*xorm.Engine) error) (err error) {
if err = InitEngine(ctx); err != nil {
return err
2014-04-05 14:46:32 +00:00
}
2015-01-22 12:49:52 +00:00
if err = x.Ping(); err != nil {
return err
}
// We have to run migrateFunc here in case the user is re-running installation on a previously created DB.
// If we do not then table schemas will be changed and there will be conflicts when the migrations run properly.
//
// Installation should only be being re-run if users want to recover an old database.
// However, we should think carefully about should we support re-install on an installed instance,
// as there may be other problems due to secret reinitialization.
if err = migrateFunc(x); err != nil {
return fmt.Errorf("migrate: %w", err)
2015-01-22 12:49:52 +00:00
}
if err = SyncAllTables(); err != nil {
return fmt.Errorf("sync database struct error: %w", err)
2014-02-19 09:50:53 +00:00
}
2015-01-23 07:54:16 +00:00
for _, initFunc := range initFuncs {
if err := initFunc(); err != nil {
return fmt.Errorf("initFunc failed: %w", err)
}
}
return nil
2014-02-18 22:48:02 +00:00
}
2014-03-20 20:04:56 +00:00
// NamesToBean return a list of beans or an error
func NamesToBean(names ...string) ([]any, error) {
beans := []any{}
if len(names) == 0 {
beans = append(beans, tables...)
return beans, nil
}
// Need to map provided names to beans...
beanMap := make(map[string]any)
for _, bean := range tables {
beanMap[strings.ToLower(reflect.Indirect(reflect.ValueOf(bean)).Type().Name())] = bean
beanMap[strings.ToLower(x.TableName(bean))] = bean
beanMap[strings.ToLower(x.TableName(bean, true))] = bean
}
gotBean := make(map[any]bool)
for _, name := range names {
bean, ok := beanMap[strings.ToLower(strings.TrimSpace(name))]
if !ok {
return nil, fmt.Errorf("no table found that matches: %s", name)
}
if !gotBean[bean] {
beans = append(beans, bean)
gotBean[bean] = true
}
}
return beans, nil
}
// DumpDatabase dumps all data from database according the special database SQL syntax to file system.
func DumpDatabase(filePath, dbType string) error {
var tbs []*schemas.Table
for _, t := range tables {
t, err := x.TableInfo(t)
if err != nil {
return err
}
tbs = append(tbs, t)
}
type Version struct {
ID int64 `xorm:"pk autoincr"`
Version int64
}
t, err := x.TableInfo(&Version{})
if err != nil {
return err
}
tbs = append(tbs, t)
if len(dbType) > 0 {
return x.DumpTablesToFile(tbs, filePath, schemas.DBType(dbType))
}
return x.DumpTablesToFile(tbs, filePath)
2014-05-05 04:55:17 +00:00
}
// MaxBatchInsertSize returns the table's max batch insert size
func MaxBatchInsertSize(bean any) int {
t, err := x.TableInfo(bean)
if err != nil {
return 50
}
return 999 / len(t.ColumnsSeq())
}
// IsTableNotEmpty returns true if table has at least one record
func IsTableNotEmpty(tableName string) (bool, error) {
return x.Table(tableName).Exist()
}
// DeleteAllRecords will delete all the records of this table
func DeleteAllRecords(tableName string) error {
_, err := x.Exec(fmt.Sprintf("DELETE FROM %s", tableName))
return err
}
// GetMaxID will return max id of the table
func GetMaxID(beanOrTableName any) (maxID int64, err error) {
_, err = x.Select("MAX(id)").Table(beanOrTableName).Get(&maxID)
return maxID, err
}
func SetLogSQL(ctx context.Context, on bool) {
e := GetEngine(ctx)
if x, ok := e.(*xorm.Engine); ok {
x.ShowSQL(on)
} else if sess, ok := e.(*xorm.Session); ok {
sess.Engine().ShowSQL(on)
}
}
[GITEA] Add slow SQL query warning - Databases are one of the most important parts of Forgejo, every interaction with Forgejo uses the database in one way or another. Therefore, it is important to maintain the database and recognize when Forgejo is not doing well with the database. Forgejo already has the option to log *every* SQL query along with its execution time, but monitoring becomes impractical for larger instances and takes up unnecessary storage in the logs. - Add a QoL enhancement that allows instance administrators to specify a threshold value beyond which query execution time is logged as a warning in the xorm logger. The default value is a conservative five seconds to avoid this becoming a source of spam in the logs. - The use case for this patch is that with an instance the size of Codeberg, monitoring SQL logs is not very fruitful and most of them are uninteresting. Recently, in the context of persistent deadlock issues (https://codeberg.org/forgejo/forgejo/issues/220), I have noticed that certain queries hold locks on tables like comment and issue for several seconds. This patch helps to identify which queries these are and when they happen. - Added unit test. (cherry picked from commit 24bbe7886fb4cb9a38c8dab8c44f4c9cbfa25481) (cherry picked from commit 6e29145b3c1455498531593d38e6a914941a12cb) (cherry picked from commit 63731e30712872bd2395eb3cf36d9996e5793645) (cherry picked from commit 3ce1a097369c132654de70df707b867e47bd1c40) (cherry picked from commit a64426907de788cc0937a7a2b16af4d2f26f7fe6)
2023-08-18 02:39:23 +00:00
type SlowQueryHook struct {
Treshold time.Duration
Logger log.Logger
}
var _ contexts.Hook = &SlowQueryHook{}
func (SlowQueryHook) BeforeProcess(c *contexts.ContextHook) (context.Context, error) {
return c.Ctx, nil
}
func (h *SlowQueryHook) AfterProcess(c *contexts.ContextHook) error {
if c.ExecuteTime >= h.Treshold {
h.Logger.Log(8, log.WARN, "[Slow SQL Query] %s %v - %v", c.SQL, c.Args, c.ExecuteTime)
}
return nil
}