CODEWITHSUNDEEP
ruby-on-railsrubyactiverecordrace-condition
grigio
grigio

Reputation: 167

find_or_create and race-condition in rails, theory and production

Hi I've this piece of code

class Place < ActiveRecord::Base
  def self.find_or_create_by_latlon(lat, lon)
    place_id = call_external_webapi
    result = Place.where(:place_id => place_id).limit(1)
    result = Place.create(:place_id => place_id, ... ) if result.empty? #!
    result
  end
end

Then I'd like to do in another model or controller

p = Post.new
p.place = Place.find_or_create_by_latlon(XXXXX, YYYYY) # race-condition
p.save

But Place.find_or_create_by_latlon takes too much time to get the data if the action executed is create and sometimes in production p.place is nil.

How can I force to wait for the response before execute p.save ? thanks for your advices

Upvotes: 0

Views: 1251

Answers (3)

Shane E.
Shane E.

Reputation: 85

I hit this inside a sidekick job that retries and gets the error repeatedly and eventually clears itself. The best explanation I've found is on a blog post here. The gist is that postgres keeps an internally stored value for incrementing the primary key that gets messed up somehow. This rings true for me because I'm setting the primary key and not just using an incremented value so that's likely how this cropped up. The solution from the comments in the link above appears to be to call ActiveRecord::Base.connection.reset_pk_sequence!(table_name) This cleared up the issue for me.

begin
   result = Place.where(:place_id => place_id).limit(1)
   result = Place.create(:place_id => place_id, ... ) if result.empty? #!
rescue ActiveRecord::StatementInvalid => error
   @save_retry_count =  (@save_retry_count || 1)
   ActiveRecord::Base.connection.reset_pk_sequence!(:place)
   retry if( (@save_retry_count -= 1) >= 0 )
   raise error
end

Upvotes: 0

Venkat D.
Venkat D.

Reputation: 3019

I had to deal with a similar problem. In our backend a user is is created from a token if the user doesn't exist. AFTER a user record is already created, a slow API call gets sent to update the users information.

def self.find_or_create_by_facebook_id(facebook_id)
  User.find_by_facebook_id(facebook_id) || User.create(facebook_id: facebook_id)
rescue ActiveRecord::RecordNotUnique => e
  User.find_by_facebook_id(facebook_id)
end

def self.find_by_token(token)
  facebook_id = get_facebook_id_from_token(token)

  user = User.find_or_create_by_facebook_id(facebook_id)

  if user.unregistered?
    user.update_profile_from_facebook
    user.mark_as_registered
    user.save
  end

  return user
end

The step of the strategy is to first remove the slow API call (in my case update_profile_from_facebook) from the create method. Because the operation takes so long, you are significantly increasing the chance of duplicate insert operations when you include the operation as part of the call to create.

The second step is to add a unique constraint to your database column to ensure duplicates aren't created.

The final step is to create a function that will catch the RecordNotUnique exception in the rare case where duplicate insert operations are sent to the database.

This may not be the most elegant solution but it worked for us.

Upvotes: 0

tadman
tadman

Reputation: 211540

You're right that this is a race condition and it can often be triggered by people who double click submit buttons on forms. What you might do is loop back if you encounter an error.

result = Place.find_by_place_id(...) ||
  Place.create(...) ||
  Place.find_by_place_id(...)

There are more elegant ways of doing this, but the basic method is here.

Upvotes: 1

Related Questions