On Sat, Jan 16, 1999 at 11:49:50AM -0800, Tom Brown wrote:
> On Sat, 16 Jan 1999, Tim Fletcher wrote:
>
> > > Now, my question is:
> > > Am I right in guessing that the problem is stale arp table entries, and is
> > > there any way to tell the rest of the net/routers that this arp entry
> > > should be flushed?
> >
> > humm have you tried setting the hardware address of the alias when you
> > bring it up to be neither of the machines and then moving it and the alias
> > over to the other machine when you move it.
>
> Last time I checked, you had to DOWN the interface before you were allowed
> to change the MAC address...
Why is that, anyway?
> this would be fine if he was talking about a 2nd card in each machine...
> but certainly isn't something you want to do if it's the only interface,
> and you aren't on the console :-(
Realistically, if your eth0 interface is down for 1/100th of a second it
won't disrupt traffic by much -- since you're doing a failover anyway, you
already have much more serious problems.
> > > I've cheched rfc826 and as far as I can see, the arp protocol makes
> > > it possible if I can generate an arp request with the .176 number as
> > > sender address, but I'm not sure if there's a nice utility for doing
> > > just that.
>
> I'm curious about the answer to the original question, because the guys
> controlling a local cisco seem to think that they can broadcast arp
> updates, yet I'm fairly sure that my linux boxes lose a few minutes of
> connectivity when the main outgoing router changes MAC addresses...
I'd like to hear the answer to that one too. I suspect it depends on the
implementation of ARP in a given OS, and we can't depend on any particular
behaviour.
Have fun,
Avery
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
