What causes "update question failed: %w" in apache/answer?

Check the wrapped driver error for lock timeout vs missing column

update question failed: %w

Error message

update question failed: %w

What it means

Wraps the xorm Update error when persisting a question's recomputed answer_count column during the v13 migration. The row-level UPDATE on the question table failed; the migration logs the offending item and aborts so the failure is visible.

Source

Thrown at internal/migrations/v13.go:165

		_, ok := questionAnswerCount[answer.QuestionID]
		if !ok {
			questionAnswerCount[answer.QuestionID] = 1
		} else {
			questionAnswerCount[answer.QuestionID]++
		}
	}
	questionList := make([]QuestionV13, 0)
	err = x.Context(ctx).Find(&questionList, &QuestionV13{})
	if err != nil {
		return fmt.Errorf("get questions failed: %w", err)
	}
	for _, item := range questionList {
		_, ok := questionAnswerCount[item.ID]
		if ok {
			item.AnswerCount = questionAnswerCount[item.ID]
			if _, err = x.Context(ctx).Cols("answer_count").Update(item, &QuestionV13{ID: item.ID}); err != nil {
				log.Errorf("update %+v config failed: %s", item, err)
				return fmt.Errorf("update question failed: %w", err)
			}
		}
	}

	return nil
}

// updateTagCount update tag count
func updateTagCount(ctx context.Context, x *xorm.Engine) error {
	tagRelList := make([]entity.TagRel, 0)
	err := x.Context(ctx).Find(&tagRelList, &entity.TagRel{})
	if err != nil {
		return fmt.Errorf("get tag rel failed: %w", err)
	}
	questionIDs := make([]string, 0)
	questionsAvailableMap := make(map[string]bool)
	questionsHideMap := make(map[string]bool)
	for _, item := range tagRelList {

View on GitHub (pinned to 3b9f137061)

Solutions

Check the wrapped driver error for lock timeout vs missing column
Confirm answer_count column exists (run the v13 DDL first)
Retry the migration after connection is restored; already-updated rows are safe to re-update
Run migrations during a maintenance window to avoid concurrent write contention

Example fix

// before
if _, err = x.Context(ctx).Cols("answer_count").Update(item, &QuestionV13{ID: item.ID}); err != nil {
    return fmt.Errorf("update question failed: %w", err)
}
// after
// ensure column exists before loop
if has, _ := x.Context(ctx).IsColumnExist("question", "answer_count"); !has {
    return fmt.Errorf("column answer_count missing; run schema migration first")
}

Defensive patterns

Strategy: retry

Validate before calling

if has, _ := x.Context(ctx).IsColumnExist("question", "answer_count"); !has {
    return fmt.Errorf("answer_count column missing; apply schema change first")
}

Try / catch

for attempt := 0; attempt < 3; attempt++ {
    if _, err = x.Context(ctx).Cols("answer_count").Update(item, &QuestionV13{ID: item.ID}); err == nil {
        break
    }
    time.Sleep(time.Duration(attempt+1) * time.Second)
}

Prevention

Run schema DDL before data backfill in the same migration
Schedule migrations in maintenance windows to avoid lock contention
Set reasonable innodb_lock_wait_timeout / statement timeouts
Make update loops idempotent so retries are safe

When it happens

Trigger: x.Context(ctx).Cols("answer_count").Update(item, &QuestionV13{ID: item.ID}) failing because the connection dropped mid-loop, answer_count column does not exist (schema not yet migrated), a lock/timeout on the row, or context cancellation.

Common situations: Migrating a large question table over a flaky connection; running the code before the v13 schema change added answer_count; deadlocks with concurrent writers to the question table.

Related errors

AI-assisted analysis of apache/answer@3b9f137061 (2026-09-05). Data as JSON: /api/errors/051961b926222722. Report an issue: GitHub.