freebsd stable - Nov 2008 - Possible regression in ifconfig under7.0 - removes valid default route

I believe there "may be" a regression in the behaviour of ifconfig or
possibly just something I've never experienced before.

Basically when changing the IP of one of our machines, it suddenly
became inaccessible. After some investigation it turned out the machine
was inaccessible from anything other than the local VLAN and continued
diagnostics determined that the process of changing the IP had also
removed the default route.

This was clearly unexpected behaviour as the new IP was on the same
VLAN as the old IP and hence no routing table updates should be
required.

I don't have any older machines to test this on but I believe we have
done this procedure in the past without any such issues so wanted to
raise it here to see if anyone else has had experience with it.

Even if this isn't a regression it may well be something worth fixing
as its quite unexpected behaviour which could render a machine totally
inaccessible.

    Regards
    Steve

===============================================This e.mail is private and
confidential between Multiplay (UK) Ltd. and the person or entity to whom it is
addressed. In the event of misdirection, the recipient is prohibited from using,
copying, printing or otherwise disseminating it or any information contained in
it.

In the event of misdirection, illegible or incomplete transmission please
telephone +44 845 868 1337
or return the E.mail to postmaster@multiplay.co.uk.

On Mon, Nov 17, 2008 at 07:11:50PM -0000, Steven Hartland
wrote:
> I believe there "may be" a regression in the behaviour of
ifconfig or
> possibly just something I've never experienced before.
>
> Basically when changing the IP of one of our machines, it suddenly
> became inaccessible. After some investigation it turned out the machine
> was inaccessible from anything other than the local VLAN and continued
> diagnostics determined that the process of changing the IP had also
> removed the default route.
>
> This was clearly unexpected behaviour as the new IP was on the same
> VLAN as the old IP and hence no routing table updates should be
> required.
>
> I don't have any older machines to test this on but I believe we have
> done this procedure in the past without any such issues so wanted to
> raise it here to see if anyone else has had experience with it.
>
> Even if this isn't a regression it may well be something worth fixing
> as its quite unexpected behaviour which could render a machine totally
> inaccessible.
I can try to reproduce this if you'd like, on a VMware session.

One case where the default route could be (I haven't checked) removed:
some NIC drivers in the past, when changing the "primary IP" (first IP
shown in ifconfig for an interface), would actually bring down the
entire interface (e.g. ifconfig XXX down) then back up again a second or
two later.  dmesg on your machine would be able to confirm whether or
not this is happening.

-- 
| Jeremy Chadwick                                jdc at parodius.com |
| Parodius Networking                       http://www.parodius.com/ |
| UNIX Systems Administrator                  Mountain View, CA, USA |
| Making life hard for others since 1977.              PGP: 4BD6C0CB |

This is a bug in my mind, but it's not a regression.  FreeBSD has done  
this for at least 10 years now.  If you are changing the IP of an  
interface, you *must* do a semicolon chained command to a route add  
default.  It's been true for as long as I can remember.

On Nov 17, 2008, at 11:11 AM, Steven Hartland wrote:
> I believe there "may be" a regression in the behaviour of
ifconfig or
> possibly just something I've never experienced before.
>
> Basically when changing the IP of one of our machines, it suddenly
> became inaccessible. After some investigation it turned out the  
> machine
> was inaccessible from anything other than the local VLAN and continued
> diagnostics determined that the process of changing the IP had also
> removed the default route.
>
> This was clearly unexpected behaviour as the new IP was on the same
> VLAN as the old IP and hence no routing table updates should be
> required.
>
> I don't have any older machines to test this on but I believe we have
> done this procedure in the past without any such issues so wanted to
> raise it here to see if anyone else has had experience with it.
>
> Even if this isn't a regression it may well be something worth fixing
> as its quite unexpected behaviour which could render a machine totally
> inaccessible.
>
>   Regards
>   Steve
>
> ===============================================> This e.mail is private
and confidential between Multiplay (UK) Ltd.
> and the person or entity to whom it is addressed. In the event of  
> misdirection, the recipient is prohibited from using, copying,  
> printing or otherwise disseminating it or any information contained  
> in it.
> In the event of misdirection, illegible or incomplete transmission  
> please telephone +44 845 868 1337
> or return the E.mail to postmaster@multiplay.co.uk.
>
> _______________________________________________
> freebsd-stable@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-stable
> To unsubscribe, send any mail to
"freebsd-stable-unsubscribe@freebsd.org
> "
-- 
Jo Rhett
Net Consonance : consonant endings by net philanthropy, open source  
and other randomness