Suppressing and Skipping Callbacks

+15 Mana ✨
Active Record Callbacks - Skipping(DOCUMENTATION)

Introduction

Sometimes you need to update the database without triggering callbacks — bulk imports, counter updates, or migrations that would fire thousands of side effects. Rails provides several methods to bypass callbacks, each with different trade-offs.

Key Concepts

  • update_column / update_columns: Update columns directly via SQL, skipping all callbacks and validations.
  • insert_all / upsert_all: Bulk insert or upsert without instantiating models.
  • delete / delete_all: Remove records via direct SQL, skipping callbacks.
  • touch: Updates updated_at with a single SQL statement, skipping validations.

Real World Context

A data migration updating 10 million rows should not trigger after_save callbacks that send emails. Counter caches need direct SQL to avoid infinite callback loops. Bulk CSV imports need insert_all for performance.

Deep Dive

update_column and update_columns

These issue a direct SQL UPDATE, bypassing the lifecycle:

ruby
article = Article.find(1)
article.update_column(:views_count, article.views_count + 1)
article.update_columns(views_count: 100, last_viewed_at: Time.current)

No callbacks, validations, or updated_at touch occurs.

insert_all and upsert_all

Bulk operations bypass models entirely:

ruby
Article.insert_all([
  { title: "Post 1", body: "Content", created_at: Time.current, updated_at: Time.current },
  { title: "Post 2", body: "Content", created_at: Time.current, updated_at: Time.current }
])

Article.upsert_all(
  [{ id: 1, title: "Updated" }],
  unique_by: :id
)

These generate a single SQL statement, orders of magnitude faster than individual creates.

Methods That Skip vs. Honor Callbacks

ruby
# SKIP callbacks:
article.update_column(:title, "New")  # Direct SQL
Article.insert_all([...])              # Bulk SQL
Article.delete_all                     # Direct DELETE
article.delete                         # Direct DELETE

# HONOR callbacks:
article.update(title: "New")           # Full lifecycle
article.save                           # Full lifecycle
article.destroy                        # Runs destroy callbacks
Article.destroy_all                    # Instantiates each, runs callbacks

The key distinction: methods that bypass object instantiation skip callbacks.

Common Pitfalls

  • Using update_column for regular updates: It skips validations. Only use when you explicitly need to bypass the lifecycle.
  • Forgetting updated_at with update_columns: These do not automatically touch updated_at. Include it manually if needed.
  • delete_all vs destroy_all: delete_all skips callbacks. destroy_all honors them.

Best Practices

  • Use insert_all / upsert_all for bulk data imports.
  • Use update_column only for performance-critical single-column updates.
  • Prefer destroy over delete unless you specifically need to skip callbacks.

Summary

  • update_column / update_columns bypass all callbacks and validations.
  • insert_all / upsert_all perform bulk operations without instantiating models.
  • delete / delete_all skip callbacks; destroy / destroy_all honor them.
  • Use callback-skipping methods for performance-critical operations.
  • Always be intentional about which side effects you are bypassing.

Code Examples

ruby
# Direct SQL update — skips callbacks and validations
article.update_column(:views_count, article.views_count + 1)

# Bulk insert — no models, no callbacks
Article.insert_all([
  { title: "Post 1", body: "Content", created_at: Time.current, updated_at: Time.current },
  { title: "Post 2", body: "Content", created_at: Time.current, updated_at: Time.current }
])

# delete skips callbacks, destroy honors them
article.delete    # Direct DELETE SQL
article.destroy   # Runs before/after_destroy callbacks
✓ Completed