On Tue, 2005-06-21 at 11:44, Caitlin Bestler wrote: > Expanding slightly on Hal's response, the correct method to correlate > an *IP* address with a device is to consult the *IP* routing tables.
> While there are exceptions where you want to control the local > address, With IBAT there is a way to force the local interface if that is what is desired. > it is far more common that you need to reach a remote > address (or be reachable from a remote address), as pointed out > in Roland's response. So the general algorithm is to check the > local routing tables to see what device would be the first hop. That's the basic algorithm which IBAT (and SDP) use. -- Hal > >From there it is merely a matter of connecting the device > declared for purposes of IP routing with the device that provides > RDMA services. > > As for the differences between *IP* and *IB* addresses, that > boils down to a determination of who is responsible for reconciling > the difference: the network stack or the application. > > > > On 6/20/05, Kevin Reilly <[EMAIL PROTECTED]> wrote: > > > > > > > > > > Maybe somebody could help me understand the proper way to map between an IP > > address assigned to a port to the > > "device name" and "port number" in the gen2 architecture. If I have an IP > > address can I map it to a name that i get back > > from ibv_get_device_name() or pass to ibv_open_device(). > > Here is the problem. Let's say there is a subsystem that used > > heartbeating over IP interfaces to ascertain the adapters > > and ports state, up or down. The data actively maintained be this > > heartbeating subsystem is used by a job scheduler that > > desires to only schedule tasks on HCAs that are UP and have active ports. > > The job scheduler does not want to launch jobs > > on adapters that are broken, not connected to a switch or not configured. > > The job scheduler needs a way to map an IP address into a name and > > port number sitable to be passed to the job to > > use when ibv_open_device() is called or to match the output of > > ibv_get_device_name(). > > > > Kevin J. Reilly > > STSM, HPC Architecture > > -Federation/HPS Chief Engineer > > -HPC interconnect architect > > (office) 845-433-7976 (tieline) 8-293-7976 > > > > _______________________________________________ > > openib-general mailing list > > [email protected] > > http://openib.org/mailman/listinfo/openib-general > > > > To unsubscribe, please visit > > http://openib.org/mailman/listinfo/openib-general > > > _______________________________________________ > openib-general mailing list > [email protected] > http://openib.org/mailman/listinfo/openib-general > > To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
