James Carlson wrote: > Sowmini.Varadhan at Sun.COM writes: > >> 3. provide per-interface prop support in ipadm and leave the "global" >> setting alone (In addition to complicating our kernel tremendously, >> I don't even see how this would work. E.g. if the "prop" >> is some thing like a rate-limit parameter, and the value is different >> for each interface, then what would the global value be?) >> > [...] > >> None of these options is ideal, but #2 seems to make the best of >> a bad-deal. Thoughts? >> > > I suggest a fifth possibility with two variants: > > 5. Leave the global variable alone. It becomes the 'default' for > newly-plumbed interfaces. Interfaces can be configured away from > that value, but the global doesn't reflect those changes. > > [..] > > b. The interface dynamically uses the global value as long as the > local value is not configured specifically for that interface. > Once it's configured, the connection with the global value is > severed. In the same way how 'ip_forwarding' ndd tunable is handled today, right?. You can set it for all the interfaces using 'ndd' and control per interface forwarding ability using 'ifconfig <inf> [-/+]router'.
Now the question is should we change 'ndd -get' output to display per interface property just like how we do for 'ip_forwarding'. For 'ndd' always meant global and there wasn't no per interface stuff. How much effort we need to put on a tool (read 'ndd') which is eventually going to die. The larger issue is how do we handle things where we will have two tools 'ndd' and 'ipadm' to begin with and people start using these tools interchangeably. ~Girish
