playframework 2.0 - Async Request in (Scala) Play "Hangs" -

i've written application (scala) play , when run on local machine, works expect. method i'm working on (there may others issue, haven't checked extensively) method triggered post request. returns async object because evaluation done method returns future.

but method behaves quite differently when deploy it. i've been chasing down day, i've gotten point have no explanation going on.

the method looks this:

  def add = action(parse.json) { request =>     async {       val req = request.body.asopt[taskrequest] map { r =>         val job = createjob(r)         submitjob(job, r)         job map { ij =>           alloworigin(created(ij).withheaders("location" -> routes.jobs.read(ij._id.stringify).url));         }       }       req.get     }   } 

overall, pretty simple. createjob method returns future[job]. submitjob call kind of "fire , forget" call trigger background stuff (that i'm not interested in waiting for). process job result produce result object when job has been created.

all works described above on local machine. when deploy this, odd happens. submitjob call triggers long running computation. said, i'm not interested in waiting why don't use future returns anywhere.

but here's odd part...on development machine, response post request (which directly calls add) comes job created. on deployment machine, response after submitjob task completed! in other words, i've got req future seemingly unrelated submitjob call (it doesn't depend on outcome), in deployment environment, async block won't return until submitjob complete (even though req fulfilled immediately).

what heck going on here? i've put println statements through code. gets past req.get call immediately. overall (http) response waits until submitjob done. async somehow know submitjob call? if so, why behave differently on 2 different machines?

any ideas?

ok, figured out. oh pain was. issue "fire , forget" task long running task. thread pool size on deployment machine ended being smaller development machine. result, long running task dropped in thread pool , clogged up. processing of asyncresult presumably preempted long running calculation (it blocking on result long running calculation, bottom line blocked).

after reading nice explanation of futures in scala 2.10 heather miller, noticed blocking construct can used temporarily expand size of thread pool single future. wrapped blocking code blocking call , did trick. presumably, allocated additional temporary worker handle blocking call and, result, asyncresult stuff got processed in timely manner.

(note, wrapping submitjob call useless because that, itself, didn't block. had go submitjob call , find place actual blocking , wrap that.)