feat: 🦭 MariaDB support 🦭 (#1724)

* feat: support MariaDB 11.4 and 11.8

gh-ost mostly works against MariaDB unchanged; this makes the few
MariaDB-specific spots correct and runs the replica test suite against
MariaDB 11.4 and 11.8 (LTS) alongside the MySQL/Percona flavors.

Core:
- ReplicaTermFor keeps the legacy slave/master replication terminology on
  MariaDB. MariaDB reports versions >= 10 (greater than the 8.4 cutoff) but
  never adopted the new replica/source terms, and its SHOW REPLICA STATUS
  still emits legacy column names, so the MySQL 8.4 term map doesn't apply.
- Accept BINLOG MONITOR as well as REPLICATION CLIENT in the grant check
  (MariaDB 10.5+ renamed the privilege).

Test harness (localtests):
- test.sh: pick slave/Master terminology for MariaDB, wait for replica
  catch-up via MASTER_GTID_WAIT/gtid_binlog_pos, and skip gtid_mode-gated
  tests where @@gtid_mode is unsupported instead of aborting the run.
- Resolve the per-flavor config dir by specificity (docker/<flavor>-<major.minor>,
  then <flavor>-<major>, then bare <flavor>) so versions that share one config
  live in a single dir. All MariaDB versions use one docker/mariadb dir (log-bin
  + ROW + log-slave-updates, native-auth users, MariaDB-GTID replication,
  MYSQL_CLIENT=mariadb since 11.x dropped the mysql-named client); a version (or
  major line) needing different config overrides it by adding its own dir. The
  compose healthcheck honours MYSQL_CLIENT.
- Make the zero-date failure tests deterministic across flavors with an
  explicit sql_mode (NO_ZERO_IN_DATE,NO_ZERO_DATE,...); they relied on MySQL's
  default sql_mode, which MariaDB doesn't share.
- Drop the incidental NOT NULL from the generated-column tests (MariaDB
  forbids NOT NULL on generated columns) so they run on MariaDB; keep
  generated-columns-unique skipped (it puts a generated column in the PRIMARY
  KEY, which MariaDB forbids, Error 1903).
- Add mariadb:11.4.12 and mariadb:11.8.8 to the CI matrix.

* feat: support MariaDB 10.5, 10.6 and 10.11

Add the MariaDB 10.x LTS lines (10.5, 10.6, 10.11) to the replica test suite.

MariaDB versions below 11.1 need one core fix: gh-ost passes transaction_isolation as
a DSN param, which the driver applies as `SET transaction_isolation=<v>` on each
connection. That system variable does not exist on MariaDB < 11.1 (which has
only tx_isolation), so the connection fails with Error 1193 "Unknown system
variable 'transaction_isolation'". tx_isolation was in turn removed in MySQL
8.0, so no single variable name is portable across the supported servers.

- mysql.OpenDB strips the transaction_isolation param and instead runs the
  SQL-standard `SET SESSION TRANSACTION ISOLATION LEVEL <level>` on each new
  connection via a connector wrapper, accepted by every supported MySQL and
  MariaDB version. GetDBUri still emits the param unchanged, so DSN
  construction and its tests are unaffected.
- Add mariadb:10.5.29, mariadb:10.6.27 and mariadb:10.11.18 to the CI matrix.

Author: conf

.github/workflows/replica-tests.yml

@@ -10,7 +10,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        image: ['mysql:5.7.41','mysql:8.0.41','mysql:8.4.3','percona/percona-server:8.0.41-32']
+        image: ['mysql:5.7.41','mysql:8.0.41','mysql:8.4.3','percona/percona-server:8.0.41-32','mariadb:10.5.29','mariadb:10.6.27','mariadb:10.11.18','mariadb:11.4.12','mariadb:11.8.8']
     env:
       TEST_MYSQL_IMAGE: ${{ matrix.image }}
 

go/logic/applier.go

@@ -174,7 +174,7 @@ func (apl *Applier) AcquireMigrationLock(ctx context.Context) error {
 	// Use a dedicated *sql.DB so the pinned connection does not consume a
 	// slot in apl.db's small pool (mysql.MaxDBPoolConnections).
 	lockURI := apl.connectionConfig.GetDBUri(apl.migrationContext.DatabaseName)
-	lockDB, err := gosql.Open("mysql", lockURI)
+	lockDB, err := mysql.OpenDB(lockURI)
 	if err != nil {
 		return fmt.Errorf("failed to open migration lock DB: %w", err)
 	}

go/logic/inspect.go

@@ -244,7 +244,8 @@ func (isp *Inspector) validateGrants() error {
 			if strings.Contains(grant, `SUPER`) && strings.Contains(grant, ` ON *.*`) {
 				foundSuper = true
 			}
-			if strings.Contains(grant, `REPLICATION CLIENT`) && strings.Contains(grant, ` ON *.*`) {
+			// MariaDB renamed REPLICATION CLIENT to BINLOG MONITOR (10.5+); accept either.
+			if (strings.Contains(grant, `REPLICATION CLIENT`) || strings.Contains(grant, `BINLOG MONITOR`)) && strings.Contains(grant, ` ON *.*`) {
 				foundReplicationClient = true
 			}
 			if strings.Contains(grant, `REPLICATION SLAVE`) && strings.Contains(grant, ` ON *.*`) {

go/mysql/replica_terminology_map.go

@@ -1,6 +1,8 @@
 package mysql
 
 import (
+	"strings"
+
 	version "github.com/hashicorp/go-version"
 )
 
@@ -25,6 +27,13 @@ var MysqlReplicaTermMap = map[string]string{
 }
 
 func ReplicaTermFor(mysqlVersion string, term string) string {
+	// MariaDB reports versions >= 10, which compare greater than the 8.4
+	// cutoff, but it never adopted the new replica/source terminology. Keep
+	// the legacy terms for it.
+	if strings.Contains(strings.ToLower(mysqlVersion), "mariadb") {
+		return term
+	}
+
 	vs, err := version.NewVersion(mysqlVersion)
 	if err != nil {
 		// default to returning the same term if we cannot determine the version

go/mysql/replica_terminology_map_test.go

@@ -0,0 +1,24 @@
+/*
+   Copyright 2025 GitHub Inc.
+	 See https://github.com/github/gh-ost/blob/master/LICENSE
+*/
+
+package mysql
+
+import "testing"
+
+func TestReplicaTermForMariaDB11KeepsLegacyTerms(t *testing.T) {
+	tests := []string{
+		"master status",
+		"slave status",
+		"Slave_IO_Running",
+		"Seconds_Behind_Master",
+	}
+
+	version := "11.4.8-MariaDB-ubu2404-log"
+	for _, term := range tests {
+		if actual := ReplicaTermFor(version, term); actual != term {
+			t.Fatalf("term %q: expected MariaDB to keep legacy term %q, got %q", term, term, actual)
+		}
+	}
+}

go/mysql/utils.go

@@ -6,14 +6,17 @@
 package mysql
 
 import (
+	"context"
 	gosql "database/sql"
+	"database/sql/driver"
 	"fmt"
 	"strings"
 	"sync"
 	"time"
 
 	"github.com/github/gh-ost/go/sql"
 
+	gomysql "github.com/go-sql-driver/mysql"
 	"github.com/openark/golib/log"
 	"github.com/openark/golib/sqlutils"
 )
@@ -48,14 +51,68 @@ func (rlg *ReplicationLagResult) HasLag() bool {
 var knownDBs map[string]*gosql.DB = make(map[string]*gosql.DB)
 var knownDBsMutex = &sync.Mutex{}
 
+// initConnector wraps a driver.Connector to run a fixed set of statements on
+// every newly established connection (e.g. setting the transaction isolation
+// level), which the DSN-param mechanism can't express portably.
+type initConnector struct {
+	driver.Connector
+	statements []string
+}
+
+func (c *initConnector) Connect(ctx context.Context) (driver.Conn, error) {
+	conn, err := c.Connector.Connect(ctx)
+	if err != nil {
+		return nil, err
+	}
+	execer, ok := conn.(driver.ExecerContext)
+	if !ok {
+		conn.Close()
+		return nil, fmt.Errorf("mysql: driver connection does not implement driver.ExecerContext")
+	}
+	for _, stmt := range c.statements {
+		if _, err := execer.ExecContext(ctx, stmt, nil); err != nil {
+			conn.Close()
+			return nil, err
+		}
+	}
+	return conn, nil
+}
+
+// OpenDB opens a MySQL connection pool for the given DSN. A transaction_isolation
+// param is applied via the SQL-standard "SET SESSION TRANSACTION ISOLATION LEVEL"
+// statement on each new connection rather than being passed to the driver as a
+// system variable:
+// - "transaction_isolation" doesn't exist on MariaDB < 11.1, while
+// - "tx_isolation" doesn't exist on MySQL 8.0+ anymore, so no single variable name is portable.
+// The standard statement is accepted by every supported MySQL and MariaDB version.
+func OpenDB(mysql_uri string) (*gosql.DB, error) {
+	cfg, err := gomysql.ParseDSN(mysql_uri)
+	if err != nil {
+		return nil, err
+	}
+	var statements []string
+	if iso := strings.Trim(cfg.Params["transaction_isolation"], `"`); iso != "" {
+		delete(cfg.Params, "transaction_isolation")
+		statements = append(statements, "SET SESSION TRANSACTION ISOLATION LEVEL "+strings.ReplaceAll(iso, "-", " "))
+	}
+	connector, err := gomysql.NewConnector(cfg)
+	if err != nil {
+		return nil, err
+	}
+	if len(statements) > 0 {
+		connector = &initConnector{Connector: connector, statements: statements}
+	}
+	return gosql.OpenDB(connector), nil
+}
+
 func GetDB(migrationUuid string, mysql_uri string) (db *gosql.DB, exists bool, err error) {
 	cacheKey := migrationUuid + ":" + mysql_uri
 
 	knownDBsMutex.Lock()
 	defer knownDBsMutex.Unlock()
 
 	if db, exists = knownDBs[cacheKey]; !exists {
-		db, err = gosql.Open("mysql", mysql_uri)
+		db, err = OpenDB(mysql_uri)
 		if err != nil {
 			return nil, false, err
 		}
@@ -88,7 +145,7 @@ func GetReplicationLagFromSlaveStatus(dbVersion string, informationSchemaDb *gos
 func GetMasterKeyFromSlaveStatus(dbVersion string, connectionConfig *ConnectionConfig) (masterKey *InstanceKey, err error) {
 	currentUri := connectionConfig.GetDBUri("information_schema")
 	// This function is only called once, okay to not have a cached connection pool
-	db, err := gosql.Open("mysql", currentUri)
+	db, err := OpenDB(currentUri)
 	if err != nil {
 		return nil, err
 	}

localtests/fail-datetime-with-zero/sql_mode

@@ -0,0 +1 @@
+STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION

localtests/fail-existing-datetime-with-zero/sql_mode

@@ -0,0 +1 @@
+STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION

localtests/generated-columns-add/extra_args

@@ -1 +1 @@
---alter="add column sum_ab int as (a + b) virtual not null"
+--alter="add column sum_ab int as (a + b) virtual"

localtests/generated-columns-rename/create.sql

@@ -3,7 +3,7 @@ create table gh_ost_test (
   id int auto_increment,
   a int not null,
   b int not null,
-  sum_ab int as (a + b) virtual not null,
+  sum_ab int as (a + b) virtual,
   primary key(id)
 ) auto_increment=1;