SUMMARY: Network Card Increment / Counter ... Curiosity

From: Brett Thorson <>
Date: Tue Apr 24 2001 - 20:03:30 EDT
The answer to this little question came from several people and I thank you
very much.
The sneaky file that should be taken out back and carefully beaten is called

This is the file that relates the devices to their appropriate names and

Beware this file as it can muck up your system, alas it can also solve your
problem much as it did this evening.  Thank you gentlemen for all your

Gurus: D Evans; D Dunham; M Baldenegro


The remnants are in /etc/path_to_inst.  That's what ties a physical
device to a particular instance.

I try not to muck around in there too much.  Have a backup before you

If you *really* want to nuke everything, (after your backup), then
1) remove the links in /dev
2) mv /etc/path_to_inst aside
3) reboot, but at the '(b) or (i) prompt', do a 'b -a'

It should prompt for all sorts of things.  When it gets to the
'path_to_inst' file, give it '/dev/null'.  It should prompt to rebuild
from scratch.  Do so.

This could potentially renumber anything on your system (ethernets, scsi

Good luck.
You've probably got a confused system at the moment. Short a full
reinstall the only advice I can give is dangerous but may help.

First you need to "unplumb" the device. This may be all you need
to do. If that doesn't get rid of it then you need the riskier path of
having to remove the entry in /devices and /etc/path_to_inst.
The path_to_inst will help you map the device to the /device

Then try a reconfigure boot, and don't forget to plumb for the new
device. (ifconfig plumb).
> => From: "Brett Thorson" <>
> => To: <>
> => Subject: QUESTION: Network Card Increment / Counter ... Curiosity
> => Date: Tue, 24 Apr 2001 15:25:44 -0400
> =>
> => This is kind of a strange question, but one that I am dying to figure
> => Here is the scenario....
> =>
> => I had two Intel network cards plugged into my Solaris box.  They were
> => different flavors, so I am guessing they used the same driver, just two
> => different instances.  Anyway.
> =>
> => Type A card was iprb0
> => Type B card was iprb1
> =>
> => That was all cool.  Then I got in some more Type B cards, and I wanted
> => make all the systems similar, so I ripped out TypeA and put in a TypeB.
> => Well at first, the system was just pissed off at me for having done
> => something so horrible.  So I took out both cards, pressed escape at
boot up,
> => made sure they had disappeared.  I had the /reconfigure in there, so it
> => would re-do everything.
> =>
> => Then what I noticed was that, it had installed the new Type B card as
> =>
> => Well I thought, this is just a remnant from the old system.  So I tried
> => everything in my power to get the new card to be iprb0.  I deleted all
> => iprb* in /dev
> => Then I went through the /devices/pci....  stuff and deleted all the
> => stuff in there.
> =>
> => Rebooted, and it came up the same way.  With only the one new Type B
card in
> => that one slot it came up iprb2.  Ok I thought, so all I have to do is
> => around with the file names, and make it think that iprb2 is actually
> =>
> => Well it didn't like that much either.  I made sure I had all my ducks
in a
> => row, but it still didn't like it.
> =>
> => I think I tried just about every combination of trying to get the
system to
> => forget about those original network cards, but it still wants to
> => that new card as iprb2.  I even tried sys-unconfig.  A neat utility,
but it
> => really didn't do much for me.
> =>
> => So I am guessing that somewhere out there in the maze of all these
> => there is a pointer, or a incrementer that is keeping track of how many
> => cards are in the system?  Or trying to associated a card with an iprb
> => or something?
> =>
> => My question is...
> =>
> => How do I get this system to forget about all the network cards ever
> => installed in there, and start over from iprb0?
> =>
> => Brett M. Thorson
> => Eko Systems Inc.
Received on Wed Apr 25 01:03:30 2001

This archive was generated by hypermail 2.1.8 : Wed Mar 23 2016 - 16:24:53 EDT