Hi All, I am looking at how we might define secondary network devices on machines.
Our current definitions make use of nicaliases, nichostnamesuffixes, nicnetworks, nicips and nictypes defined on a per-group basis (-eno1, -eno2, -ib0) and membership of the group defines the mpi network. How would we go about defining the secondary network when using consistent device names? I found this email from 4 years ago looking at the same issue. https://sourceforge.net/p/xcat/mailman/message/35193146/ I see that getadapter.pm exists, but I'm not sure how we would define the secondary network to allow for a large group of dissimilar device names - especially if the device name is being discovered in post? It looks like it is possible - with nicsadapter maybe? - but I can't think it through. Has anyone solved this problem in practice or have an idea on how it might be done? Cheers L. -- _______________________________________________ xCAT-user mailing list xCAT-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xcat-user
