Chewy now offers several callback mechanisms and builtin support for the most popular background processing systems. ChewyKiqqer is no longer compatible with the newest version of chewy and will no longer be maintained.
I suggest that you just roll with Chewy's defaults. Please note that for reliable operation with background queues like sidekiq you should enable the option to queue on after_commit. Otherwise your jobs may work on outdated data.
This is an alternative update/callback mechanism for Chewy. It queues the updates as Sidekiq jobs.
You can pass backrefs like with the standard chewy mechanism, but the job itself will always receive an array of ids.
It is possible to install more multiple update hooks.
Add this line to your application's Gemfile:
gem 'chewy_kiqqer'
And then execute:
$ bundle
Or install it yourself as:
$ gem install chewy_kiqqer
Just add the module and set it up:
class User < ActiveRecord::Base
include ChewyKiqqer::Mixin
async_update_index index: 'users#user', queue: :other_than_default, backref: :something
end
You can also include the mixin into ActiveRecord::Base in an initializer if it should be generally available. The queue name is optional. You can also set a default queue name for your application with:
ChewyKiqqer.default_queue = :my_queue
Giving a backref is also optional (also check the chewy docs for the concept). The backref is the element which will be indexed. The default is to use the current record.
# use the current record for the backref
... backref: :self
# call a method on the current record to get the backref
... backref: :some_method
# Pass a proc. It will be called with the current record to get the backref
... backref: -> (rec) { |rec| rec.do_something }
The kiqqer does not use Chewy's atomic update, since that functionality is deeply linked with Chewy's syncronous update mechanism.
Instead, ChewyKiqqer will bind itself to the #after_commit callback, which means that it will only trigger a new job after a complete database transaction. This behaviour also ensures that only one job is enqueued per transaction.
However, if you have multiple database transactions, the kiqqer will still queue multiple jobs. The same is true when you enqueue jobs manually.
ChewyKiqqer uses locking via redis to ensure that all updates for one database record are run sequentially. This prevents race conditions which could lead to outdated data being written to the index otherwise.
Logging is disabled by default, but you can set ChewyKiqqer.logger if you need log output (e.g. ChewyKiqqer.logger = Rails.logger). ChewyKiqqer uses ActiveSupport notifications, which you can also subscribe to.
See log_subscriber.rb for more info.
- Fork it
- Create your feature branch (
git checkout -b my-new-feature) - Commit your changes (
git commit -am 'Add some feature') - Push to the branch (
git push origin my-new-feature) - Create new Pull Request