Today on a Bull EPC400 cluster running AIX 4.3.1 I unearthed the following
entries in /dev
>ls -l /dev/bp*
> cr-------- 1 root system 56, 0 Jul 14 1999 /dev/bpf0
> cr-------- 1 root system 56, 1 Jul 14 1999 /dev/bpf1
> cr-------- 1 root system 56, 2 Jul 14 1999 /dev/bpf2
> cr-------- 1 root system 56, 3 Jul 14 1999 /dev/bpf3
Anyone recognise them, because we sure as hell don't !!!
Regards,
Nick Buckley
AIX Administrator,
NCM,
Cardiff,
Wales,
United Kingdom
e-mail: nicholas...@ncmgroup.com
(Please note: all newsgroup offerings are made in a personal
capacity and in no way are the responsibility of my employer)
p.s. Did you know that "AIX Operating System" is an anagram of "Mega sexy
partitions" !!!
>Today on a Bull EPC400 cluster running AIX 4.3.1 I unearthed the following
>entries in /dev
>
>>ls -l /dev/bp*
>> cr-------- 1 root system 56, 0 Jul 14 1999 /dev/bpf0
>> cr-------- 1 root system 56, 1 Jul 14 1999 /dev/bpf1
>> cr-------- 1 root system 56, 2 Jul 14 1999 /dev/bpf2
>> cr-------- 1 root system 56, 3 Jul 14 1999 /dev/bpf3
>
>Anyone recognise them, because we sure as hell don't !!!
These are the Berkeley Packet Filter devices used by tcpdump.
--
Jens-Uwe Mager <pgp-mailto:62CFDB25>
They're a little funny, too. They don't show up in /dev until tcpdump
has been used at least once since boot. Makes them a little annoying
if you want to use bpf for other programs such as dhcp ...
--
James Carlson, System Architect <car...@ibnets.com>
IronBridge Networks / 55 Hayden Avenue 71.246W Vox: +1 781 372 8132
Lexington MA 02421-7996 / USA 42.423N Fax: +1 781 372 8090
"PPP Design and Debugging" --- http://people.ne.mediaone.net/carlson/ppp
Yes, but from what I understand they do not want application programs to
use them. AIX has its own AF_NDD style socket mechanism that can be
used to do things like DHCP.
By coincidence, we created a VG with Major Number 56 ( same as these
devices ) on another node.
We could not synchronise HACMP because these devices were also MAJ=56.
So, we exported an re-imported the VG on its host node with a
non-conflicting MAJNO.
That solved that problem, but to end this thread, Jens/James, can you think
of a good reason why we shouild NOT remove these devices ??
And finally, they don't show up via lsdev -C, so can we just rm them ?
Nick.
Jens-Uwe Mager wrote in message ...
>By coincidence, we created a VG with Major Number 56 ( same as these
>devices ) on another node.
>
>We could not synchronise HACMP because these devices were also MAJ=56.
This looks like a bug to me, I would recommend to submit that as a bug
to your IBM support channel. There should be no way to create such a
major number conflict accidently.
>So, we exported an re-imported the VG on its host node with a
>non-conflicting MAJNO.
>
>That solved that problem, but to end this thread, Jens/James, can you think
>of a good reason why we shouild NOT remove these devices ??
>
>And finally, they don't show up via lsdev -C, so can we just rm them ?
This devices will come back as soon as someone uses the tcpdump command.
I would believe IBM should fix this so these bpf devices are first class
citizens in the AIX device configuration architecture or remove them.
I would actually prefer if they could extend the AF_NDD mechanism to
contain the equivalent of the packet filter engine, these would be much
cleaner architecture wise compared to putting it into a character device
(Linux does this, for example).
We don't support using BPF for anything except tcpdump. That's why
the devices aren't there until you run tcpdump.
-Dave
It would be nice to have a way to configure them in on start-up, since
they're useful for application programs even if no other IBM-supplied
program uses them.
Currently, I do "/usr/sbin/tcpdump -d host 0" in my /etc/rc.tcpip in
order to get them in place without actually dumping anything. That's
a little hackish ...