Add logging to long migrations (#11647)

* Add logging to long migrations

Also fix v136 to not use models

Signed-off-by: Andrew Thornton <art27@cantab.net>

models/migrations/v115.go

 	"crypto/md5"
 	"fmt"
 	"io/ioutil"
+	"math"
 	"os"
 	"path/filepath"
+	"time"
 
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
 		LowerName string `xorm:"UNIQUE NOT NULL"`
 		Avatar    string
 	}
+
+	ticker := time.NewTicker(5 * time.Second)
+	defer ticker.Stop()
+
+	count, err := x.Count(new(User))
+	if err != nil {
+		return err
+	}
+	log.Info("%d User Avatar(s) to migrate ...", count)
+
 	deleteList := make(map[string]struct{})
 	start := 0
+	migrated := 0
 	for {
 		if err := sess.Begin(); err != nil {
 			return fmt.Errorf("session.Begin: %v", err)
 			}
 
 			deleteList[filepath.Join(setting.AvatarUploadPath, oldAvatar)] = struct{}{}
+			migrated++
+			select {
+			case <-ticker.C:
+				log.Info(
+					"%d/%d (%2.0f%%) User Avatar(s) migrated (%d old avatars to be deleted) in %d batches. %d Remaining ...",
+					migrated,
+					count,
+					float64(migrated)/float64(count)*100,
+					len(deleteList),
+					int(math.Ceil(float64(migrated)/float64(50))),
+					count-int64(migrated))
+			default:
+			}
 		}
 		if err := sess.Commit(); err != nil {
 			_ = sess.Rollback()
 		}
 	}
 
+	deleteCount := len(deleteList)
+	log.Info("Deleting %d old avatars ...", deleteCount)
+	i := 0
 	for file := range deleteList {
 		if err := os.Remove(file); err != nil {
 			log.Warn("os.Remove: %v", err)
 		}
+		i++
+		select {
+		case <-ticker.C:
+			log.Info(
+				"%d/%d (%2.0f%%) Old User Avatar(s) deleted. %d Remaining ...",
+				i,
+				deleteCount,
+				float64(i)/float64(deleteCount)*100,
+				deleteCount-i)
+		default:
+		}
 	}
+
+	log.Info("Completed migrating %d User Avatar(s) and deleting %d Old Avatars", count, deleteCount)
+
 	return nil
 }
 

models/migrations/v128.go

 
 import (
 	"fmt"
+	"math"
 	"path/filepath"
 	"strings"
+	"time"
 
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
 		limit = 50
 	}
 
+	ticker := time.NewTicker(5 * time.Second)
+	defer ticker.Stop()
+
+	count, err := x.Count(new(PullRequest))
+	if err != nil {
+		return err
+	}
+	log.Info("%d Pull Request(s) to migrate ...", count)
+
 	i := 0
+	start := 0
 	for {
 		prs := make([]PullRequest, 0, 50)
-		if err := x.Limit(limit, i).Asc("id").Find(&prs); err != nil {
+		if err := x.Limit(limit, start).Asc("id").Find(&prs); err != nil {
 			return fmt.Errorf("Find: %v", err)
 		}
 		if len(prs) == 0 {
 			break
 		}
 
-		i += len(prs)
+		start += 50
 		for _, pr := range prs {
 			baseRepo := &Repository{ID: pr.BaseRepoID}
 			has, err := x.Table("repository").Get(baseRepo)
 			}
 			pr.MergeBase = strings.TrimSpace(pr.MergeBase)
 			x.ID(pr.ID).Cols("merge_base").Update(pr)
+			i++
+			select {
+			case <-ticker.C:
+				log.Info("%d/%d (%2.0f%%) Pull Request(s) migrated in %d batches. %d PRs Remaining ...", i, count, float64(i)/float64(count)*100, int(math.Ceil(float64(i)/float64(limit))), count-int64(i))
+			default:
+			}
 		}
 	}
-
+	log.Info("Completed migrating %d Pull Request(s) in: %d batches", count, int(math.Ceil(float64(i)/float64(limit))))
 	return nil
 }

models/migrations/v134.go

 
 import (
 	"fmt"
+	"math"
 	"path/filepath"
 	"strings"
+	"time"
 
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
 		limit = 50
 	}
 
+	ticker := time.NewTicker(5 * time.Second)
+	defer ticker.Stop()
+	count, err := x.Where("has_merged = ?", true).Count(new(PullRequest))
+	if err != nil {
+		return err
+	}
+	log.Info("%d Merged Pull Request(s) to migrate ...", count)
+
 	i := 0
+	start := 0
 	for {
 		prs := make([]PullRequest, 0, 50)
-		if err := x.Limit(limit, i).Asc("id").Where("has_merged = ?", true).Find(&prs); err != nil {
+		if err := x.Limit(limit, start).Asc("id").Where("has_merged = ?", true).Find(&prs); err != nil {
 			return fmt.Errorf("Find: %v", err)
 		}
 		if len(prs) == 0 {
 			break
 		}
 
-		i += len(prs)
+		start += 50
 		for _, pr := range prs {
 			baseRepo := &Repository{ID: pr.BaseRepoID}
 			has, err := x.Table("repository").Get(baseRepo)
 			}
 			pr.MergeBase = strings.TrimSpace(pr.MergeBase)
 			x.ID(pr.ID).Cols("merge_base").Update(pr)
+			i++
+			select {
+			case <-ticker.C:
+				log.Info("%d/%d (%2.0f%%) Pull Request(s) migrated in %d batches. %d PRs Remaining ...", i, count, float64(i)/float64(count)*100, int(math.Ceil(float64(i)/float64(limit))), count-int64(i))
+			default:
+			}
 		}
 	}
+
+	log.Info("Completed migrating %d Pull Request(s) in: %d batches", count, int(math.Ceil(float64(i)/float64(limit))))
 	return nil
 }

models/migrations/v136.go

 
 import (
 	"fmt"
+	"math"
+	"path/filepath"
+	"strings"
+	"time"
 
 	"code.gitea.io/gitea/models"
+	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
-	pull_service "code.gitea.io/gitea/services/pull"
 
 	"xorm.io/xorm"
 )
 
 func addCommitDivergenceToPulls(x *xorm.Engine) error {
+	type Repository struct {
+		ID        int64 `xorm:"pk autoincr"`
+		OwnerID   int64 `xorm:"UNIQUE(s) index"`
+		OwnerName string
+		LowerName string `xorm:"UNIQUE(s) INDEX NOT NULL"`
+		Name      string `xorm:"INDEX NOT NULL"`
+	}
+
+	type PullRequest struct {
+		ID int64 `xorm:"pk autoincr"`
+
+		CommitsAhead  int
+		CommitsBehind int
+
+		BaseRepoID int64 `xorm:"INDEX"`
+		BaseBranch string
+
+		HasMerged      bool   `xorm:"INDEX"`
+		MergedCommitID string `xorm:"VARCHAR(40)"`
+	}
 
 	if err := x.Sync2(new(models.PullRequest)); err != nil {
 		return fmt.Errorf("Sync2: %v", err)
 	}
 
-	var last int
+	last := 0
+	migrated := 0
+
 	batchSize := setting.Database.IterateBufferSize
 	sess := x.NewSession()
 	defer sess.Close()
+
+	ticker := time.NewTicker(5 * time.Second)
+	defer ticker.Stop()
+	count, err := sess.Where("has_merged = ?", false).Count(new(PullRequest))
+	if err != nil {
+		return err
+	}
+	log.Info("%d Unmerged Pull Request(s) to migrate ...", count)
+
 	for {
 		if err := sess.Begin(); err != nil {
 			return err
 		if len(results) == 0 {
 			break
 		}
-		last += len(results)
+		last += batchSize
 
 		for _, pr := range results {
-			divergence, err := pull_service.GetDiverging(pr)
+			baseRepo := &Repository{ID: pr.BaseRepoID}
+			has, err := x.Table("repository").Get(baseRepo)
+			if err != nil {
+				return fmt.Errorf("Unable to get base repo %d %v", pr.BaseRepoID, err)
+			}
+			if !has {
+				log.Error("Missing base repo with id %d for PR ID %d", pr.BaseRepoID, pr.ID)
+				continue
+			}
+			userPath := filepath.Join(setting.RepoRootPath, strings.ToLower(baseRepo.OwnerName))
+			repoPath := filepath.Join(userPath, strings.ToLower(baseRepo.Name)+".git")
+
+			gitRefName := fmt.Sprintf("refs/pull/%d/head", pr.Index)
+
+			divergence, err := git.GetDivergingCommits(repoPath, pr.BaseBranch, gitRefName)
 			if err != nil {
 				log.Warn("Could not recalculate Divergence for pull: %d", pr.ID)
 				pr.CommitsAhead = 0
 				pr.CommitsBehind = 0
 			}
-			if divergence != nil {
-				pr.CommitsAhead = divergence.Ahead
-				pr.CommitsBehind = divergence.Behind
-			}
+			pr.CommitsAhead = divergence.Ahead
+			pr.CommitsBehind = divergence.Behind
+
 			if _, err = sess.ID(pr.ID).Cols("commits_ahead", "commits_behind").Update(pr); err != nil {
 				return fmt.Errorf("Update Cols: %v", err)
 			}
+			migrated++
 		}
 
 		if err := sess.Commit(); err != nil {
 			return err
 		}
+		select {
+		case <-ticker.C:
+			log.Info(
+				"%d/%d (%2.0f%%) Pull Request(s) migrated in %d batches. %d PRs Remaining ...",
+				migrated,
+				count,
+				float64(migrated)/float64(count)*100,
+				int(math.Ceil(float64(migrated)/float64(batchSize))),
+				count-int64(migrated))
+		default:
+		}
 	}
+	log.Info("Completed migrating %d Pull Request(s) in: %d batches", count, int(math.Ceil(float64(migrated)/float64(batchSize))))
 	return nil
 }