[zeromq-dev] debugging performance problems

[Andrew Hume]

well, i actually have have some viewgraphs describing the flows,
but i see the note on the bottom says "proprietary restricted -- NDA required".
never mind.

i should be okay; most of the flow is push/pull and the rest is PUB/SUB.
and the main case of PUB/SUB is more or less a device where its hould be easy to count
ins and outs.

just to verify, to set hwm limits on a push pull, i set it just on the pull end,
after the zmq_socket call and before the zmq_connect call. correct?

	andrew

On May 16, 2011, at 7:04 AM, Ian Barber wrote:

> 
> 
> On Mon, May 16, 2011 at 2:54 PM, Andrew Hume <andrew at research.att.com> wrote:
> how can i tell if it was hit?
> 
> generally, the process links don't have hwm set, but some do.
> is your advice to set them on all links?
> 
> 
> Yeah, I think having HWM set across all links is a good idea. For detection, depends on the socket type - polling for blocking should work for HWM on REQ/XREQ/PUSH sockets, but I don't think there is any way to measure it for PUB/XREP as it will just drop. The only way to get detection then is to have sequence numbers and some kind of nack, which seems like a lot of work for the problem here - you're getting to the point where you could implement as a series of request-response connections and track that way, which is a pretty big model change if you are using pub sub. 
> 
> Ian
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev


------------------
Andrew Hume  (best -> Telework) +1 623-551-2845
andrew at research.att.com  (Work) +1 973-236-2014
AT&T Labs - Research; member of USENIX and LOPSA




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20110516/e51f83f2/attachment.htm>