Re: ActiveSupport::Notifications threading question

165 views
Skip to first unread message

Frederick Cheung

unread,
Sep 12, 2012, 8:00:57 AM9/12/12
to rubyonra...@googlegroups.com


On Wednesday, September 12, 2012 2:11:41 AM UTC+1, Jeffrey Jones wrote:
Hello all.

I am planning on using ActiveSupport::Notifications and have a question about threading. For the purposes of this question I am assuming config.threadsafe! is enabled.

Will a subscriber execute in the same thread as the code that caused it to be triggered or do they run in separate threads?

I would have thought that was easy to test. Currently everything happens on the thread triggering the event. There's also code in rails (eg ActiveRecord's logsubscriber) that use thread local variables in notification processing so I wouldn't that would be changed in a hurry.

Fred

 
The reason I ask is that I would like to do something like

In Controller Code:

def action
  Module.events[Thread.current] = Event.new # Module.events returns a hash
  # rest of action code
end

And elsewhere:

ActiveSupport::Notifications.subscribe "process_action.action_controller" do |name, start, finish, id, payload|
  event = Module.events[Thread.current] # returns the Event that was created in the controller.
  # Do processing on event
end

Obviously if the Subscriber runs in a different thread then this will not work since Thread.current will not match so I wanted to check. Note that in this case I do not want to pass the Event as part of the payload because I want to be able to use existing rails built-in notifications without having to modify them.

Cheers

Jeff




Jeffrey Jones

unread,
Sep 12, 2012, 8:06:07 PM9/12/12
to rubyonra...@googlegroups.com
I have done various testing using MRI and JRuby with config.threadsafe! enabled and I THINK that it does always execute in the same thread.

However since the log files are interleaved when dealing with many requests and logs written from the subscriber do not have access to the request object for logging with a tag using something like uuid I cannot be 100% sure which is why I was hoping someone would be able to give a definitive answer.

cheers

Jeff
--
You received this message because you are subscribed to the Google Groups "Ruby on Rails: Talk" group.
To post to this group, send email to rubyonra...@googlegroups.com.
To unsubscribe from this group, send email to rubyonrails-ta...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msg/rubyonrails-talk/-/zaXAeHVURIUJ.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Jeffrey Jones

unread,
Sep 12, 2012, 8:27:33 PM9/12/12
to rubyonra...@googlegroups.com
Sorry I was talking out of my arse in the last post. Correct version:

I have done various testing using MRI and JRuby with config.threadsafe!  and it appears to work in all the tests I have thrown at it, but that gives me 99% certainty (I could have missed soemthing).

I tried tracing through the code on github but quickly got lost which is why I asked the question here to see if someone (who does know the code) could give a definitive "Yes" or "no"

Cheers

Jeff

Frederick Cheung

unread,
Sep 13, 2012, 9:54:51 AM9/13/12
to rubyonra...@googlegroups.com


On Thursday, September 13, 2012 1:28:11 AM UTC+1, Jeffrey Jones wrote:
Sorry I was talking out of my arse in the last post. Correct version:

I have done various testing using MRI and JRuby with config.threadsafe!  and it appears to work in all the tests I have thrown at it, but that gives me 99% certainty (I could have missed soemthing).

I tried tracing through the code on github but quickly got lost which is why I asked the question here to see if someone (who does know the code) could give a definitive "Yes" or "no"

 
And of course there's the difference between something just happening to behave in a certain way with the current implementation and something which a documented part of the interface. There's not much in the way of documentation around this stuff but given that existing rails internals assume that the delivery thread is the same as the thread processing the notification I'd say it was pretty safe to rely on.


Fred
Reply all
Reply to author
Forward
0 new messages