How do you skip failed migrations? (rake db:migrate)

Question

I can't seem to find an option or anything that allows me to skip migrations.

I know what you're thinking: "you should never have to do that..."

I need to skip a migration that makes changes to specific user records that don't exist in my development database. I don't want to change the migration because it is not part of the source I am supposed to be working with. Is there a way to skip a migration or skip failed migrations?

Thanks in advance!

Answer 1

I think you should fix the offending migrations to be less fragile, I'd guess that a couple of if statements and perhaps a rescue would be sufficient.

But, if fixing the migrations really isn't an option, you can fake it in various ways. First of all, you could just comment out the migration methods, run rake db:migrate, and then uncomment (or revert) the offending migration.

You can also fake it inside the database but this sort of chicanery is not recommended unless you know what you're doing and you don't mind manually patching things up when you (inevitably) make a mistake. There is a table in your database called schema_migrations that has a single varchar(255) column called version; this table is used by db:migrate to keep track of which migrations have been applied. All you need to do is INSERT the appropriate version value and rake db:migrate will think that the migration has been done. Find the offending migration file:

db/migrate/99999999999999_XXXX.rb

then go into your database and say:

insert into schema_migrations (version) values ('99999999999999');

where 99999999999999 is, of course, the number from the migration's file name. Then running rake db:migrate should skip that migration.

I'd go with the second option before the third, I'm only including the "hack schema_versions" option for completeness.

Answer 2

This is a good way to do it for one-off errors.

db:migrate:up VERSION=my_version

This will run one specific migration's "up" actions. (There is also the opposite if you need it, just replace "up" with "down".) So this way you can either run the future migration that makes the older one (that you need to skip) work, or just run each migration ahead of it selectively.

I also believe that you can redo migrations this way:

rake db:migrate:redo VERSION=my_version

I have not tried that method personally, so YMMV.

Answer 3

I had an issue where I had a migration to add a table that already existed, so in my case I had to skip this migration as well, because I was getting the error

SQLite3::SQLException: table "posts" already exists: CREATE TABLE "posts"

I simply commented out the content of the create table method, ran the migration, and then uncommented it out. It's kind of a manual way to get around it, but it worked. See below:

class CreatePosts < ActiveRecord::Migration
  def change
    # create_table :posts do |t|
    #   t.string :title
    #   t.text :message
    #   t.string :attachment
    #   t.integer :user_id
    #   t.boolean :comment
    #   t.integer :phase_id

    #   t.timestamps
    # end
  end
end

Answer 4

If you have to do that, your app's migrations are messed up!

Inserts all missing migrations:

def insert(xxx)
  ActiveRecord::Base.connection.execute("insert into schema_migrations (version) values (#{xxx})") rescue nil
end

files = Dir.glob("db/migrate/*")
files.collect { |f| f.split("/").last.split("_").first }.map { |n| insert(n) }