*** From dhcp-server -- To unsubscribe, see the end of this message. ***
-----------
For Alan:
I had a problem with the dhcpd and we came to the following conclusions
about the linux bridge...
-----------
On Mon, 26 Apr 1999, Ted Lemon wrote:
> *** From dhcp-server -- To unsubscribe, see the end of this message. ***
>
> > Till this point it works... but whenever a client on eth0 issues
> > a dhcprequest, it gets bridged eth1 catches it to dhcpd, dhcpd sends it
> > out on eth1.... and it's never to be seen on the eth0 segment...
>
> Hm. I would argue that this is incorrect behaviour on the part of the
> Linux kernel. Either the kernel should bridge all packets, or none.
> It shouldn't be possible to transmit a packet on one half of a bridged
> network, and the Linux kernel needs to ensure that this is the case.
Problem is that they call it a bridge and it acts like a switch...
What they should do now is check the destination mac addr and transmit it
onto the correct device or both...
>
> The way this should work is that if you have two ethernet interfaces
> that are going to be bridged together, the act of bridging them
> together should create a third, virtual ethernet interface, and that
> third interface is what should get ifconfig'd. The two bridged
> interfaces shouldn't even appear to a user application to be running,
> much less to have IP addresses.
Yep that's the solution... even for the switch (which it's now)...
This also avoids problems with samba etc not knowing on which side to
broadcast etc... it is the same network... :)
>
> All this Sunday-morning quarterbacking is of course probably of
> limited comfort to you. Unfortunately, the amount of hacking required
> to teach the DHCP server to work around this problem is pretty
> significant, and I'm reluctant to do it for such an obscure case,
> particularly when I think the API Linux is presenting is the real
> problem - it should ensure that this never happens. For example, you
> should never have had to specify eth1 on the command line.
As you stated above it should be implemented with a virtual device.
>
> The usual response on the part of the Linux networking crowd when I
> act all curmudgeonly about stuff like this is to insist that if I want
> it fixed, I should bloody well fix it myself - perfectly reasonable.
> Unfortunately, right now, my pain level is low, so I'm probably not
> going to this. If you want it fixed, you will probably have to do
> some hacking. If I start to get the sense that a lot of Linux people
> are losing because of this, though, I might change my tune, so your
> other alternative is to try to attract consensus on that... :')
Well I'll at least put it on my todo list.. but it's quite lengthy...
I'll first have to "finish" /dev/changer and maybe I'll take some time
for this then... it would sure mean a lot less configging of my box as it
would look like one network then...
>
> _MelloN_
>
>
>
> ------------------------------------------------------------------------------
> To unsubscribe from this list, please visit http://www.fugue.com/dhcp/lists
> If you are without web access, or if you are having trouble with the web page,
> please send mail to [EMAIL PROTECTED] Please try to use the web
> page first - it will take a long time for your request to be processed by hand.
>
> Archives for this mailing list are available at
>http://www.webnology.com/list-archives/dhcp/dhcp-server
>
> ------------------------------------------------------------------------------
>
------------------------------------------------------------------------------
To unsubscribe from this list, please visit http://www.fugue.com/dhcp/lists
If you are without web access, or if you are having trouble with the web page,
please send mail to [EMAIL PROTECTED] Please try to use the web
page first - it will take a long time for your request to be processed by hand.
Archives for this mailing list are available at
http://www.webnology.com/list-archives/dhcp/dhcp-server
------------------------------------------------------------------------------
