From: Stephen Hemminger <shemminger@osdl.org>
To: "Jørgen Hovland" <jorgen@hovland.cx>
Cc: bridge@lists.osdl.org
Subject: Re: [Bridge] physical interface on a bridge
Date: Tue, 21 Feb 2006 14:17:43 -0800 [thread overview]
Message-ID: <20060221141743.32d66879@localhost.localdomain> (raw)
In-Reply-To: <01f701c63728$6203bef0$4a27b3d5@tungemaskin>
On Tue, 21 Feb 2006 21:50:00 +0100
Jørgen Hovland <jorgen@hovland.cx> wrote:
> Hi
>
> Is there a way to either:
> Find the real ifindex/ifname a mac-address is bound to
> or
> Find the real ifindex/ifname of an incoming packet
> ?
>
> I am writing a dhcp server and need to know what real interface the dhcp request packet came from. An acceptable solution would be to get the interface by the mac-address, but that can be faked so I would rather get the interface by knowing where the data actually came from. Data is IP, UDP broadcast.
> I _could_ use raw sockets. The problem is when I do that, the program is using ~8% cpu on a 3.2ghz xeon64 just reading packets without doing anything due to the amount of traffic passing through the box (~200mbit and increasing) so that doesn't look like a good idea.
Why should the app care. If forwarding database is working correctly, the source mac
of the incoming packet will be in the list and any response to it will go out that interface.
> brctl showmacs returns a list of port numbers, but they dont make much sense to me. They do not seem to be in the same order I added the interfaces? Is there a mapping here?
>
> Example,
> jorgen@ams41:/$ /tmp/brctl showmacs test0
> port no mac addr is local? ageing timer
> 2 00:04:e2:a8:3b:d7 no 0.24
> 1 00:08:a1:85:39:fd no 17.31
> 133 00:0d:88:a3:61:4a no 9.90
> 1 00:14:22:b0:cd:e0 yes 0.00
> 133 00:16:c7:f5:8f:e2 no 0.48
>
> Port 133 is the 901'th interface (0x385) I added to bridge test0. What does 133 point to? The ifindex of this physical interface is 912 (0x390) (retrieved with SIOCGIFINDEX).
Arbitrary index assigned by bridge for STP usage. Slots get reused as ports are deleted and added.
>
> Secondly,
> I seem to be unable to add more than around 1024 interfaces to a single bridge. Is there a way to increase this limit?
Increase BR_PORT_BITS (you can go up to 15) but you will lose priority bits on the spanning tree.
Also, why? You performance is going to start to fall off with so many interfaces. Can't you
partition to multiple machines?
next prev parent reply other threads:[~2006-02-21 22:17 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-02-21 20:50 [Bridge] physical interface on a bridge Jørgen Hovland
2006-02-21 22:17 ` Stephen Hemminger [this message]
2006-02-22 8:42 ` Jørgen Hovland
2006-02-22 16:02 ` Stephen Hemminger
2006-02-23 8:30 ` Jørgen Hovland
2006-02-23 8:39 ` Jørgen Hovland
2006-02-23 18:27 ` Stephen Hemminger
2006-02-24 13:20 ` Jørgen Hovland
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20060221141743.32d66879@localhost.localdomain \
--to=shemminger@osdl.org \
--cc=bridge@lists.osdl.org \
--cc=jorgen@hovland.cx \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.