Posted: Thu Feb 05, 2009 8:06 am Post subject: Strange TimeoutControl node performance issue
Newbie
Joined: 05 Feb 2009 Posts: 6
Hello all
I'm experiencing very poor performance from a TimeoutControl node. From the time the previous node passes in the message till the following node after the timeout control processes the message is >500ms. I believe I am passing the correct TimeoutRequest format because the TimeoutNotification does in fact work.
I'm setting InputLocalEnvironment.TimeoutRequest to:
The TimeoutNotification node with corresponding UniqueId is triggered after 5 seconds. I just don't understand why TimeoutControl takes 500ms to process this message, it can't possible take that long to write a local non-persistant message to a queue can it?
Am I failing to grasp something basic here? Hoping someone can shed some light.
P.S. Anything else relvant I can think of:
broker Timeout Q depth is currently 0
after i remove TimeoutControl node total time is reduced back to original <10ms
Joined: 12 Jan 2006 Posts: 1494 Location: Singapore
I guess its waiting for 5 sec to propagate the first message initially.. i.e. taken from Interval value... i.e. broker takes it as send a message after every 5 seconds of the start time.... and hence its not sending the message at 0
If you want the message to be propagated initially(and only once) then give as below
I guess its waiting for 5 sec to propagate the first message initially.. i.e. taken from Interval value... i.e. broker takes it as send a message after every 5 seconds of the start time.... and hence its not sending the message at 0
If you want the message to be propagated initially(and only once) then give as below
I do not want the message propogated initially, maybe I didn't make myself clear.
Simple test flow I've made to illustrate the problem goes like this:
Flow1: HttpIn -> JavaComputNode(Create TimeoutRequest) -> TimeoutControl -> Trace -> HttpReply
Flow2: TimeoutNotification -> Trace
I would like the TimeoutControl node to start 5second timer, and immediatley continue onto the Trace and HttpReply nodes. It currently takes around 500ms to start said timer before passing control to the trace node. This makes no sense to me, how would this behavior be expected? The interval of 5 is specified in seconds not ms anyway, so it does not appear it is waiting the interval time either.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum