feat: gh-ost 完整策略引擎 — 13项配置+环境差异+云数据库适配+DDL智能分析

配置项(13项): - ghost_enabled: 全局开关(默认关闭) - ghost_default_mode: 默认执行模式(dry_run/execute) - ghost_osc_priority: OSC优先级(native_first/ghost_first/ghost_only) - ghost_env_mode: 环境模式(development/production) - ghost_cloud_type: 云数据库类型(self_hosted/aliyun_rds/polardb/aws_aurora) - ghost_auto_trigger_rows: 自动触发行数阈值(万) - ghost_auto_trigger_ddl: 自动触发DDL类型 - ghost_path/chunk_size/max_load/critical_load/max_lag_millis/allow_on_master 新增API: - GET /ghost/config: 查看全部gh-ost配置 - POST /ghost/analyze: DDL智能分析(解析DDL类型+行数检查+策略推荐) - 云数据库: --aliyun-rds / --aws 自动适配策略逻辑: - native_first + development → 推荐Native DDL - native_first + production + DDL匹配 → 推荐gh-ost - ghost_first/ghost_only → 强制gh-ost - 行数超阈值 → 额外推荐gh-ost
2026-05-27 00:22:33 +08:00
1 parent 5d59ca427e
commit b2b13661fd
@@ -28,6 +28,8 @@ type GhOstTask struct {
 	Execute        bool   `json:"execute"`
 	ExactRowCount  bool   `json:"exact_row_count"`
 	PostponeCutOver bool  `json:"postpone_cut_over"`
+	AliyunRDS      bool   `json:"aliyun_rds,omitempty"`  // --aliyun-rds
+	AWSAurora      bool   `json:"aws_aurora,omitempty"`  // --aws
 }

 // GhOstResult holds the execution result
@@ -96,6 +98,13 @@ func RunGhOst(ctx context.Context, task GhOstTask, ghOstPath string) (*GhOstResu
 	if task.PostponeCutOver {
 		args = append(args, "--postpone-cut-over-flag-file=/tmp/gh-ost-cutover.flag")
 	}
+	// Cloud-specific flags
+	if task.AliyunRDS {
+		args = append(args, "--aliyun-rds")
+	}
+	if task.AWSAurora {
+		args = append(args, "--aws")
+	}

 	// Dry-Run vs Execute
 	// gh-ost defaults to noop (dry-run), only --execute actually applies changes
@@ -235,6 +244,13 @@ func buildArgs(task GhOstTask, ghOstPath string) []string {
 	if task.PostponeCutOver {
 		args = append(args, "--postpone-cut-over-flag-file=/tmp/gh-ost-cutover.flag")
 	}
+	// Cloud-specific flags
+	if task.AliyunRDS {
+		args = append(args, "--aliyun-rds")
+	}
+	if task.AWSAurora {
+		args = append(args, "--aws")
+	}

 	// gh-ost defaults to noop, only --execute actually applies changes
 	if task.Execute {
@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"net/http"
 	"strconv"
+	"strings"
 	"time"

 	"sql-platform/internal/ghost"
@@ -12,6 +13,8 @@ import (
 	"github.com/gin-gonic/gin"
 )

+// ==================== Ghost Settings Helpers ====================
+
 // getSetting reads a system_setting value by key
 func getSetting(key, fallback string) string {
 	var s model.SystemSetting
@@ -23,7 +26,7 @@ func getSetting(key, fallback string) string {

 // isGhostEnabled checks if gh-ost is enabled in system settings
 func isGhostEnabled() bool {
-	return getSetting("ghost_enabled", "true") == "true"
+	return getSetting("ghost_enabled", "false") == "true"
 }

 // getGhostPath returns the gh-ost binary path from settings
@@ -31,6 +34,180 @@ func getGhostPath() string {
 	return getSetting("ghost_path", "gh-ost")
 }

+// getGhostConfig returns all ghost-related settings as a map for API responses
+func getGhostConfig() gin.H {
+	return gin.H{
+		"enabled":              isGhostEnabled(),
+		"default_mode":         getSetting("ghost_default_mode", "dry_run"),
+		"osc_priority":         getSetting("ghost_osc_priority", "native_first"),
+		"env_mode":             getSetting("ghost_env_mode", "development"),
+		"cloud_type":           getSetting("ghost_cloud_type", "self_hosted"),
+		"auto_trigger_rows":    getSettingInt("ghost_auto_trigger_rows", 10),
+		"auto_trigger_ddl":     getSetting("ghost_auto_trigger_ddl", "ADD COLUMN,MODIFY COLUMN,ADD INDEX"),
+		"chunk_size":           getSettingInt("ghost_default_chunk_size", 1000),
+		"max_load":             getSetting("ghost_max_load", "Threads_running=25"),
+		"critical_load":        getSetting("ghost_critical_load", "Threads_running=50"),
+		"allow_on_master":      getSetting("ghost_allow_on_master", "true") == "true",
+		"max_lag_millis":       getSettingInt("ghost_max_lag_millis", 1500),
+	}
+}
+
+// getSettingInt reads a system_setting as integer
+func getSettingInt(key string, fallback int) int {
+	if v, err := strconv.Atoi(getSetting(key, "")); err == nil {
+		return v
+	}
+	return fallback
+}
+
+// ==================== Ghost OSC Strategy Engine ====================
+
+// DDLAnalysis contains the analysis result of an ALTER statement
+type DDLAnalysis struct {
+	DDLTypes     []string // extracted DDL operation types
+	NeedsOSC     bool     // whether OSC tool is recommended
+	Reason       string   // explanation for the recommendation
+	RecommendedTool string // "native", "gh-ost", "pt-osc"
+}
+
+// analyzeDDL inspects the ALTER statement to determine DDL types and OSC need
+func analyzeDDL(alterStmt string) DDLAnalysis {
+	upper := strings.ToUpper(alterStmt)
+	var types []string
+
+	ddls := map[string]string{
+		"ADD COLUMN": "ADD COLUMN", "DROP COLUMN": "DROP COLUMN",
+		"MODIFY COLUMN": "MODIFY COLUMN", "CHANGE COLUMN": "CHANGE COLUMN",
+		"ADD INDEX": "ADD INDEX", "DROP INDEX": "DROP INDEX",
+		"ADD UNIQUE": "ADD INDEX", "ADD PRIMARY KEY": "ADD INDEX",
+		"DROP PRIMARY KEY": "DROP INDEX",
+		"RENAME COLUMN": "CHANGE COLUMN", "ALGORITHM": "ALGORITHM",
+	}
+	for pattern, ddlType := range ddls {
+		if strings.Contains(upper, pattern) {
+			types = appendIfMissing(types, ddlType)
+		}
+	}
+
+	// Determine recommendation based on config
+	oscPriority := getSetting("ghost_osc_priority", "native_first")
+	envMode := getSetting("ghost_env_mode", "development")
+
+	analysis := DDLAnalysis{DDLTypes: types, RecommendedTool: "native"}
+
+	switch oscPriority {
+	case "ghost_only":
+		analysis.NeedsOSC = true
+		analysis.RecommendedTool = "gh-ost"
+		analysis.Reason = "平台配置为仅使用 gh-ost 模式"
+	case "ghost_first":
+		analysis.NeedsOSC = true
+		analysis.RecommendedTool = "gh-ost"
+		analysis.Reason = "平台配置为优先使用 gh-ost"
+	default: // native_first
+		if envMode == "production" {
+			// Production: check auto-trigger conditions
+			triggerDDLs := strings.Split(getSetting("ghost_auto_trigger_ddl", ""), ",")
+			for _, t := range types {
+				for _, trigger := range triggerDDLs {
+					if strings.TrimSpace(trigger) == t {
+						analysis.NeedsOSC = true
+						analysis.RecommendedTool = "gh-ost"
+						analysis.Reason = fmt.Sprintf("生产环境 + DDL类型[%s]匹配自动触发规则", t)
+						break
+					}
+				}
+			}
+		} else {
+			analysis.Reason = "开发/测试环境，建议使用 Native DDL"
+		}
+	}
+
+	return analysis
+}
+
+// appendIfMissing adds a string to slice only if not present
+func appendIfMissing(slice []string, s string) []string {
+	for _, el := range slice {
+		if el == s {
+			return slice
+		}
+	}
+	return append(slice, s)
+}
+
+// ==================== Ghost API Handlers ====================
+
+// GET /api/v1/ghost/config — return current ghost configuration
+func GhOstGetConfig(c *gin.Context) {
+	c.JSON(http.StatusOK, getGhostConfig())
+}
+
+// POST /api/v1/ghost/analyze — analyze DDL and return OSC recommendation
+func GhOstAnalyzeDDL(c *gin.Context) {
+	var req struct {
+		InstanceID int    `json:"instance_id"`
+		Database   string `json:"database"`
+		Table      string `json:"table"`
+		Alter      string `json:"alter"`
+	}
+	if err := c.ShouldBindJSON(&req); err != nil {
+		c.JSON(http.StatusBadRequest, gin.H{"detail": err.Error()})
+		return
+	}
+
+	analysis := analyzeDDL(req.Alter)
+
+	// Also check table row count if instance_id provided
+	rowCount := -1
+	if req.InstanceID > 0 {
+		rowCount = getTableRowCount(req.InstanceID, req.Database, req.Table)
+	}
+
+	// Check auto-trigger by row count
+	triggerRows := getSettingInt("ghost_auto_trigger_rows", 10) * 10000
+	if rowCount >= 0 && rowCount > triggerRows && !analysis.NeedsOSC {
+		analysis.NeedsOSC = true
+		analysis.RecommendedTool = "gh-ost"
+		analysis.Reason = fmt.Sprintf("表行数(%d)超过阈值(%d)，推荐使用 gh-ost", rowCount, triggerRows)
+	}
+
+	c.JSON(http.StatusOK, gin.H{
+		"enabled":          isGhostEnabled(),
+		"ddl_types":        analysis.DDLTypes,
+		"needs_osc":        analysis.NeedsOSC,
+		"recommended_tool": analysis.RecommendedTool,
+		"reason":           analysis.Reason,
+		"table_rows":       rowCount,
+		"trigger_rows":     triggerRows,
+		"config":           getGhostConfig(),
+	})
+}
+
+// getTableRowCount estimates row count via INFORMATION_SCHEMA
+func getTableRowCount(instanceID int, database, table string) int {
+	var inst model.Instance
+	if model.DB.First(&inst, instanceID).Error != nil {
+		return -1
+	}
+	if inst.DBType != "mysql" {
+		return -1
+	}
+
+	db, err := getDBConnection(&inst, database)
+	if err != nil {
+		return -1
+	}
+	defer db.Close()
+
+	var count int
+	row := db.QueryRow("SELECT TABLE_ROWS FROM INFORMATION_SCHEMA.TABLES WHERE TABLE_SCHEMA = ? AND TABLE_NAME = ?", database, table)
+	if err := row.Scan(&count); err != nil {
+		return -1
+	}
+	return count
+}
+
 // POST /api/v1/ghost/dry-run
 func GhOstDryRun(c *gin.Context) {
 	if !isGhostEnabled() {
@@ -66,40 +243,6 @@ func GhOstDryRun(c *gin.Context) {
 	c.JSON(http.StatusOK, result)
 }

-// buildGhOstTask resolves instance_id to connection params
-func buildGhOstTask(instanceID int, database, table, alter string) (*ghost.GhOstTask, error) {
-	var inst model.Instance
-	if err := model.DB.First(&inst, instanceID).Error; err != nil {
-		return nil, fmt.Errorf("instance %d not found", instanceID)
-	}
-	if inst.DBType != "mysql" {
-		return nil, fmt.Errorf("gh-ost only supports MySQL instances, got %s", inst.DBType)
-	}
-	return &ghost.GhOstTask{
-		Host:     inst.Host,
-		Port:     inst.Port,
-		User:     inst.Username,
-		Password: inst.Password,
-		Database: database,
-		Table:    table,
-		Alter:    alter,
-	}, nil
-}
-
-// applyGhostSettings applies system_settings to the task (chunk-size, max-load, cloud flags)
-func applyGhostSettings(task *ghost.GhOstTask) {
-	// Chunk size
-	if cs, err := strconv.Atoi(getSetting("ghost_default_chunk_size", "1000")); err == nil && cs > 0 {
-		task.ChunkSize = cs
-	}
-	// Max load
-	if ml := getSetting("ghost_max_load", ""); ml != "" {
-		task.MaxLoad = ml
-	}
-	// Allow on master
-	task.AllowOnMaster = getSetting("ghost_allow_on_master", "true") == "true"
-}
-
 // POST /api/v1/ghost/execute
 func GhOstExecute(c *gin.Context) {
 	if !isGhostEnabled() {
@@ -112,6 +255,7 @@ func GhOstExecute(c *gin.Context) {
 		Database   string `json:"database"`
 		Table      string `json:"table"`
 		Alter      string `json:"alter"`
+		Force      bool   `json:"force"` // skip auto-trigger check
 	}
 	if err := c.ShouldBindJSON(&req); err != nil {
 		c.JSON(http.StatusBadRequest, gin.H{"detail": err.Error()})
@@ -134,6 +278,62 @@ func GhOstExecute(c *gin.Context) {
 	})
 }

+// ==================== Ghost Helpers ====================
+
+// buildGhOstTask resolves instance_id to connection params
+func buildGhOstTask(instanceID int, database, table, alter string) (*ghost.GhOstTask, error) {
+	var inst model.Instance
+	if err := model.DB.First(&inst, instanceID).Error; err != nil {
+		return nil, fmt.Errorf("instance %d not found", instanceID)
+	}
+	if inst.DBType != "mysql" {
+		return nil, fmt.Errorf("gh-ost only supports MySQL instances, got %s", inst.DBType)
+	}
+
+	task := &ghost.GhOstTask{
+		Host:     inst.Host,
+		Port:     inst.Port,
+		User:     inst.Username,
+		Password: inst.Password,
+		Database: database,
+		Table:    table,
+		Alter:    alter,
+	}
+
+	// Apply cloud-specific flags
+	cloudType := getSetting("ghost_cloud_type", "self_hosted")
+	switch cloudType {
+	case "aliyun_rds":
+		task.AliyunRDS = true
+	case "aws_aurora":
+		task.AWSAurora = true
+	}
+
+	return task, nil
+}
+
+// applyGhostSettings applies all system_settings to the task
+func applyGhostSettings(task *ghost.GhOstTask) {
+	// Chunk size
+	if cs := getSettingInt("ghost_default_chunk_size", 1000); cs > 0 {
+		task.ChunkSize = cs
+	}
+	// Max load
+	if ml := getSetting("ghost_max_load", ""); ml != "" {
+		task.MaxLoad = ml
+	}
+	// Critical load
+	if cl := getSetting("ghost_critical_load", ""); cl != "" {
+		task.CriticalLoad = cl
+	}
+	// Max lag millis
+	if ml := getSettingInt("ghost_max_lag_millis", 1500); ml > 0 {
+		task.MaxLagMillis = ml
+	}
+	// Allow on master
+	task.AllowOnMaster = getSetting("ghost_allow_on_master", "true") == "true"
+}
+
 // GET /api/v1/ghost/status/:taskId
 func GhOstStatus(c *gin.Context) {
 	taskID := c.Param("taskId")