Sowmini.Varadhan at Sun.COM writes: > On (03/10/09 13:21), Paul Durrant wrote: > > Why would it be a lie? There is a prop. caller old_icmp_err_rate_limit > > and you can print its value. However, neither bge0 not bge1 use it. Why > > is that a problem? ndd is marked as deprecated, so surely it's 'user > > beware'? > > catching up on this thread (which progressed a bit while I wasn't looking)- > I see what you/Jim mean: basically redefine ndd variables as "values that > would be configured unless they are over-ridden by per-interface specific > props from ipadm" > > And it wouldn't be too hard to implement Jim's (b) suggestion- we'd > have to track a dont_use_ndd_global_default bit per interface, per property. > > Yes, that is feasible, and probably better than 1-4 in my list.
One consideration is what to do with "ndd -set" on the global variable. With ip_forwarding, in addition to the global flag used for subsequent new interfaces, we also go out and blast all of the existing interfaces to the new setting. That makes the design more like my 5(a): the interface just has a simple copy of the value created at plumb time, and you can set it on a per-interface basis if you want. I dunno if that's better or worse than 5(b). It might be close to the noise. -- James Carlson, Solaris Networking <james.d.carlson at sun.com> Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677
