I really do appreciate your help so far, so thanks again for clearing things up. One question to help clarifying things. Why are we seeing a non linear growth in execution time with these "wide" concurrent iterators? Do you think splitting these into smaller batches can make a difference?
On Thursday, April 24, 2014 1:26:32 PM UTC+3, John Mettraux wrote: > > > On Thu, Apr 24, 2014 at 02:02:04AM -0700, Dan Kilman wrote: > > > > Following your advice, we are going to try the following: > > 1. execute a workflow for each node > > > > 2. try and minimize the workitem to the bear minimum and fetch data only > on > > demand. > > Hello, > > unfortunately, about this point 2, I think you already did for the best by > using merge-type="ignore". Of course, if the workitem is super fat, it'll > save time to shrink it. > > > Did you happen to run the test case we supplied? Is there anything you > can > > suggest to futher investigate? > > No, I'm sorry, we're in the middle of a "going live with the product" > session > in the dayjob. These days I only have 1 hour that I can dedicate to open > source work per day. > > > We know about noisy mode but we are heaving some hard time interpreting > > what is going on as there are many subprocesses and many different > branches > > and its very hard to follow. > > I don't think noisy mode we'll help. As said in my previous email, ruote > was > not meant to go "wide" like this. It simply does its job and that's too > slow > for you. No bug it seems, just a bad performance surprise. > > That's a reason why I don't dive into running your example. > > Best regards, > > John > > -- -- 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 --- You received this message because you are subscribed to the Google Groups "ruote" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
