> > However, in the common case with e.g. two interfaces together in an IPMP > > group that has DHCP data addresses, when the group fails, the IPMP IP > > interface's IFF_RUNNING flag will be cleared and thus its routes removed > > IFF_RUNNING is traditionally used to indicate that the resources needed > by the driver are available/allocated. Is it correct to turn off this > flag on the ipmpN interface when the under-interfaces have failed?
The semantics of IFF_RUNNING actually match up pretty well. For instance, consider the case where all the interfaces in the group fail and then one repairs: the restoration of IFF_RUNNING triggers DAD to run again, which is the right thing to do. It also causes DHCP to recheck its leases, which is also proper. Also, ignoring the issue with route-based targets, it enables higher-level redundancy -- e.g., consider a system with four interfaces grouped into two IPMP groups with dynamic routing between them. I have a proposed solution, which I will cover in another email. -- meem
