On 9 July 2013 05:53, Marten Feldtmann <[email protected]> wrote:
> Despite this problem the XP programs are doing their job - sending and > receiving the data. > This implies some form of spin-wait as per Laurent's reply. The symptoms should typically appear after the first packet loss and NAK generation. The question is how you are driving 0mq IO and whether it is the 0mq IO thread spinning. Some annotation to the runtime may be required. With PGM you can enable console level trace roles and play with different timers to see if that improves the operation. Also, one might need to verify this problem is not arising: https://groups.google.com/forum/#!topic/openpgm-dev/IW2kAe0ty-c Setting timer mechanism: https://code.google.com/p/openpgm/wiki/OpenPgm5CReferenceRunWithTheseCapabilities Setting trace log levels: https://code.google.com/p/openpgm/wiki/OpenPgm5CReferenceErrorHandling -- Steve-o
_______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
