Hello again,
I looked into reserve method as you suggested and it ended up to be perfect
place to implement affinity for workers. I also reverted storage
participant to being single threaded so worker is mopnopolized by
participants and long running participants queue up.
I patched Sequel::Storage with following code:
<<CODE
def reserve(doc)
allowed_to_process?(doc) && super
end
def allowed_to_process?(doc)
if doc['workitem'] && doc['workitem']['fields'] &&
doc['workitem']['fields']['params']
!doc['workitem']['fields']['params']['long'] ||
!REGISTRY[:worker_skip_long]
else
true
end
end
<<CODE
I created to rake tasks run_worker and run_worker_long. In second task I
set the registry key to true so it processes the long running tasks.
I know it does not look too elegant and I am monkey patching sequel
storage, but for now it works perfectly and gives me ability to call
participant and give it an affinity for long running tasks.
For example:
<<PDEF
Ruote.process_definition :name => 'test', :revision => '0.1' do
sequence do
alice :task => 'clean car'
bob :task => 'sell car', :long => true
end
end
<<PDEF
Will be processed by any worker until it hits participant bob. Once it hits
bob, only worker that does not skip long running tasks will accept that
participant.
Thanks,
Iuri
--
you received this message because you are subscribed to the "ruote users" group.
to post : send email to [email protected]
to unsubscribe : send email to [email protected]
more options : http://groups.google.com/group/openwferu-users?hl=en
