Add retry logic for instant DDL on lock wait timeout

go/logic/applier.go

@@ -305,7 +305,32 @@ func (this *Applier) AttemptInstantDDL() error {
 		return err
 	}
 	// We don't need a trx, because for instant DDL the SQL mode doesn't matter.
-	_, err := this.db.Exec(query)
+	return retryOnLockWaitTimeout(func() error {
+		_, err := this.db.Exec(query)
+		return err
+	}, this.migrationContext.Log)
+}
+
+// retryOnLockWaitTimeout retries the given operation on MySQL lock wait timeout
+// (errno 1205). Non-timeout errors return immediately. This is used for instant
+// DDL attempts where the operation may be blocked by a long-running transaction.
+func retryOnLockWaitTimeout(operation func() error, logger base.Logger) error {
+	const maxRetries = 5
+	var err error
+	for i := 0; i < maxRetries; i++ {
+		if i != 0 {
+			logger.Infof("Retrying after lock wait timeout (attempt %d/%d)", i+1, maxRetries)
+			RetrySleepFn(time.Duration(i) * 5 * time.Second)
+		}
+		err = operation()
+		if err == nil {
+			return nil
+		}
+		var mysqlErr *drivermysql.MySQLError
+		if !errors.As(err, &mysqlErr) || mysqlErr.Number != 1205 {
+			return err
+		}
+	}
 	return err
 }
 

go/logic/applier_test.go

@@ -8,9 +8,12 @@ package logic
 import (
 	"context"
 	gosql "database/sql"
+	"errors"
 	"strings"
 	"testing"
+	"time"
 
+	drivermysql "github.com/go-sql-driver/mysql"
 	"github.com/stretchr/testify/require"
 	"github.com/stretchr/testify/suite"
 
@@ -198,6 +201,71 @@ func TestApplierInstantDDL(t *testing.T) {
 	})
 }
 
+func TestRetryOnLockWaitTimeout(t *testing.T) {
+	oldRetrySleepFn := RetrySleepFn
+	defer func() { RetrySleepFn = oldRetrySleepFn }()
+	RetrySleepFn = func(d time.Duration) {} // no-op for tests
+
+	logger := base.NewMigrationContext().Log
+
+	lockWaitTimeoutErr := &drivermysql.MySQLError{Number: 1205, Message: "Lock wait timeout exceeded"}
+	nonRetryableErr := &drivermysql.MySQLError{Number: 1845, Message: "ALGORITHM=INSTANT is not supported"}
+
+	t.Run("success on first attempt", func(t *testing.T) {
+		calls := 0
+		err := retryOnLockWaitTimeout(func() error {
+			calls++
+			return nil
+		}, logger)
+		require.NoError(t, err)
+		require.Equal(t, 1, calls)
+	})
+
+	t.Run("retry on lock wait timeout then succeed", func(t *testing.T) {
+		calls := 0
+		err := retryOnLockWaitTimeout(func() error {
+			calls++
+			if calls < 3 {
+				return lockWaitTimeoutErr
+			}
+			return nil
+		}, logger)
+		require.NoError(t, err)
+		require.Equal(t, 3, calls)
+	})
+
+	t.Run("non-retryable error returns immediately", func(t *testing.T) {
+		calls := 0
+		err := retryOnLockWaitTimeout(func() error {
+			calls++
+			return nonRetryableErr
+		}, logger)
+		require.ErrorIs(t, err, nonRetryableErr)
+		require.Equal(t, 1, calls)
+	})
+
+	t.Run("non-mysql error returns immediately", func(t *testing.T) {
+		calls := 0
+		genericErr := errors.New("connection refused")
+		err := retryOnLockWaitTimeout(func() error {
+			calls++
+			return genericErr
+		}, logger)
+		require.ErrorIs(t, err, genericErr)
+		require.Equal(t, 1, calls)
+	})
+
+	t.Run("exhausts all retries", func(t *testing.T) {
+		calls := 0
+		err := retryOnLockWaitTimeout(func() error {
+			calls++
+			return lockWaitTimeoutErr
+		}, logger)
+		require.ErrorIs(t, err, lockWaitTimeoutErr)
+		require.Equal(t, 5, calls)
+	})
+}
+
 type ApplierTestSuite struct {
 	suite.Suite