On Fri, 2017-10-06 at 13:54 -0400, Patrick Talbert wrote: > > My goal with the patch is not to prevent some one from bonding > whichever interfaces they want, only to notify them that what they are > doing is *likely* to be less than ideal from a performance > perspective. Even if some theoretical load balancing bonding mode was > intelligent enough to consider NUMA when choosing a transmit > interface, it never has control over the interface traffic is received > on (excluding the strange balance-alb mode).
Note that following the NUMA node of current process would lead to reordering for TCP flows. XPS makes sure we stick to one TXQ to avoid reorders. I am not sure your patch will really help, since you do not warn if a process from another NUMA node tries to send packet to a bonding driver using slaves on another NUMA node.
