AIX using en0 instead of en1, no matter what
Brian Beuchaw
This past week, my AIX 4.3.3 system has decided that it wants to use the
en0 (built-in 10 Mbps interface) for outgoing traffic all the time
(instead of the add-in en1 100 Mbps card). It's been working just fine
for over a year, but *something* changed (I haven't done anything to that
box admin-wise for a very long time and I'm the only admin), the error
logs show nothing, and it just started happening.
I eventually figured out (with a little help from a past post to this
group) that I can do a rmdev -l en0, rmdev -l en1, mkdev -l en1, mkdev -l
en0 and it will start sending all traffic through en1, but it doesn't last
through a boot. I didn't figure it would, so I eventually detached en0,
but any attempt to use the network resulted in a "network not available"
message (sorry, I didn't write the message down and I can't recreate it as
the machine is being used in production currently, so I can't play with
it). I tried marking en0 as down, but the same message popped up.
/etc/rc.net and /etc/rc.tcpip show no ifconfigs going on. I've tried to
do a little research, but it hasn't been very productive, so I was
wondering what the silver bullet was for keeping en0 out of the mix (I
know how to get it down, but why does something keep thinking it has to
use it instead of en1?)....
Thanks in advance for any info or pointers on what command/manual to look
in....
Brian Beuchaw
--
If you want to reply to this message by mail, you will
have to change the reply address to brian....@pb.com
Michael Selle
> Hello all,
>
> This past week, my AIX 4.3.3 system has decided that it wants to use the
> en0 (built-in 10 Mbps interface) for outgoing traffic all the time
> (instead of the add-in en1 100 Mbps card). It's been working just fine
> for over a year, but *something* changed (I haven't done anything to that
> box admin-wise for a very long time and I'm the only admin), the error
> logs show nothing, and it just started happening.
So both interfaces do have the same IP configuration?
This leads to strange behavior.
Are you shure en0 is 10Mbps and en1 100?
>
> I eventually figured out (with a little help from a past post to this
> group) that I can do a rmdev -l en0, rmdev -l en1, mkdev -l en1, mkdev -l
> en0 and it will start sending all traffic through en1, but it doesn't last
> through a boot. I didn't figure it would, so I eventually detached en0,
> but any attempt to use the network resulted in a "network not available"
> message (sorry, I didn't write the message down and I can't recreate it as
After you detach an interface the routing tabel ist updated.
Configure ip on the other interface once again
check routig table
>
> the machine is being used in production currently, so I can't play with
> it). I tried marking en0 as down, but the same message popped up.
> /etc/rc.net and /etc/rc.tcpip show no ifconfigs going on. I've tried to
ODM Database stores some conifguration data
Brian or Sharon Beuchaw
> Brian Beuchaw wrote:
>
>> Hello all,
>>
>> This past week, my AIX 4.3.3 system has decided that it wants to use the
>> en0 (built-in 10 Mbps interface) for outgoing traffic all the time
>> (instead of the add-in en1 100 Mbps card). It's been working just fine
>> for over a year, but *something* changed (I haven't done anything to that
>> box admin-wise for a very long time and I'm the only admin), the error
>> logs show nothing, and it just started happening.
>
> So both interfaces do have the same IP configuration?
Yeah, I kept en0's configuration so I could go back to it without too much
trouble in case I had problems with putting the new card in
> This leads to strange behavior.
I would think it wouldn't matter if en0 is detached or down - am I wrong?
> Are you shure en0 is 10Mbps and en1 100?
Absolutely 100%
>> I eventually figured out (with a little help from a past post to this
>> group) that I can do a rmdev -l en0, rmdev -l en1, mkdev -l en1, mkdev -l
>> en0 and it will start sending all traffic through en1, but it doesn't last
>> through a boot. I didn't figure it would, so I eventually detached en0,
>> but any attempt to use the network resulted in a "network not available"
>> message (sorry, I didn't write the message down and I can't recreate it as
>
> After you detach an interface the routing tabel ist updated.
> Configure ip on the other interface once again
> check routig table
I did that and it looks fine, but when I reboot, it all goes back to en0
being the only interface.
>> the machine is being used in production currently, so I can't play with
>> it). I tried marking en0 as down, but the same message popped up.
>> /etc/rc.net and /etc/rc.tcpip show no ifconfigs going on. I've tried to
>
> ODM Database stores some conifguration data
Yep, I knew that, but how can I get it to permanently recognize en1 as the
interface to use instead of en0?
I found some info in an AIX manual (Sys Mgmt Concepts: Opsys and Devices,
chapter Unconfiguring Comm Devices) and followed it, but the same damn
thing happened. It said to do an ifconfig en0 detach, then rmdev -l ent0
and it'll be unconfigured. But this also comes back at reboot.
I also just found out that sometime between yesterday afternoon and this
morning, the interface switched back to en0 (and there are no cron jobs
running except errclear), so I'm totally confused as to why this is
happening randomly and sporadically....
Thanks for the post anyway....
brian
--
If you want to reply to this message by mail, you will
have to change the reply address to beu...@enteract.com
Paul Landay
ifconfig en0 down detach
rmdev -l en0 -d
----
or you should be able to do this to keep the values,
but but make the ineffective on reboot:
ifconfig en0 down detach
chdev -l en0 -a state=detach
Paul Landay
Brian Beuchaw
Thanks Paul,
This worked quite well. I thought I had done a chdev (or maybe I just
thought about it)... But I was doing this in the middle of things with an
HP-UX system (moving disks around), a z/OS system, and other paperwork
admin things, so I wasn't as focused as I should've been. I've got 2
Solaris 6 systems, a Solaris 8 system, an HP-UX 10.20, an HP-UX 11.0, an
AIX 4.1.5, an AIX 4.3.3, a RedHat, and a z/OS system to take care of, so
it gets a bit much at times. :-)
Thanks again,
brian
--
If you want to reply to this message by mail, you will
have to change the reply address to brian....@pb.com
Kollie
You may never use IP addresses within the same subnet on different LAN
adapters in the same AIX-box (this is for AIX4.x.x, for AIX5.x.x I'm not
sure).
Latr,
Eric
"Brian Beuchaw" <nos...@nobody.org> wrote in message
news:aeql6j$n3e$1...@bob.news.rcn.net...