On Tue, 2005-05-24 at 15:28, Sean Hefty wrote: > Hal Rosenstock wrote: > >>>DREQ --------> > >>> <------- DREP > >>>DREQ --------> > >>>DREQ --------> > >> > >>I ran some more tests, and I can't reproduce this. Can you describe your > >>test setup wrt the CM more? > > > > > > It's just running the kdapl quit test between 2 machines. > > Can you describe what this test does? This result can occur if the DREP is > lost and the sending cm_id exits timewait before the DREQ is received a > second time.
I don't think the DREP is lost and is coming back quickly (177.436 usec after the DREQ). The connection was set up for 4 retries. I see the original one followed by the DREP followed by 4 more DREQs (for that same connection). > How many QPs are connecting at once? Just one? Just one in this test invocation. -- Hal _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
