On Fri, 2007-02-09 at 12:14, Michael Arndt wrote: > Hi, > > > umad_send takes the timeout in msec. 100 msec is too short. Try > > something on the order of seconds. Note also that negative 'timeout_ms' > > value makes the kernel wait for the reply forever. > > I have tried many values, but sooner or later the umad_send broke down, > which is bad because the SM thinks a port or node is unreachable if there > didn't come an response. All works fine if I sleep after every send but that > can't be the right track. What can I do or is there a known bug in the > libibumad that I have slipped?
I have no clue; I don't really understand what you have changed so it is hard to know. > I modified the __osm_state_mgr_sweep_hop_1 function so it send not just one > packets with [0][1] but also [0][1][1], [0][1][1][1]. I don't understand what you are trying to do and the scope of your changes. > Any there it happens too that one packet is not be sent sometimes. I can't parse this sentence. > I'm wondering because if the SM get all PortInfos from an switch > there are be many sends too, but it seems be that it works. Yes, the SM will poll for each port on a switch for its PortInfo and each of these is a separate SubnGet. > Is the packet size for the umad_send 256 max? It depends on the MAD type. SMPs are limited to a single MAD (256 bytes) whereas GMPs can be larger if the class supports RMPP (as SA does). -- Hal > Thanks Michael > > > > > _______________________________________________ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general