imaging computer generates bogus ifcfg-eth1 - Mandrake

This is a discussion on imaging computer generates bogus ifcfg-eth1 - Mandrake ; Mandriva 2007. harddrake and haldaemon are turned off. I made an image of one computer onto another, and after the image was written to disk the script creating that image changed only one file: /etc/sysconfig/network which gets a new HOSTNAME ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: imaging computer generates bogus ifcfg-eth1

  1. imaging computer generates bogus ifcfg-eth1

    Mandriva 2007. harddrake and haldaemon are turned off. I made an image
    of one computer onto another, and after the image was written to disk
    the script creating that image changed only one file:

    /etc/sysconfig/network

    which gets a new HOSTNAME value but is itself otherwise unchanged.

    Then the system reboots. When it comes up it now has an

    /etc/sysconfig/network-scripts/ifcfg-eth1

    entry (set for dhcpd).

    Why??? There's only one NIC on this system. Which part of
    Mandriva 2007 took it upon itself to create this bogus ifcfg entry?
    The only clue was in /var/log/messages:

    Feb 13 04:33:15 monkey02 kernel: eth0: setting full-duplex.
    Feb 13 04:33:15 monkey02 ifplugd(eth0)[1713]: Using interface
    eth0/00:E0:81:22:BB:E3 with driver <3c59x> (version: )
    Feb 13 04:33:15 monkey02 ifplugd(eth0)[1713]: Using detection mode:
    SIOCETHTOOL
    Feb 13 04:33:15 monkey02 ifplugd(eth0)[1713]: Initialization complete,
    link beat detected.
    Feb 13 04:33:15 monkey02 ifplugd(eth0)[1713]: Executing
    '/etc/ifplugd/ifplugd.action eth0 up'.
    Feb 13 04:33:15 monkey02 udevd-event[1699]: rename_netif: error changing
    netif name eth0 to eth1: Device or resource busy

    Those messages never appeared on the node which was imaged. Since
    effectively the only thing that changed was the time stamp
    on /etc/sysconfig/network apparently that triggered this event. Who
    knows where and why?

    Thanks,

    David Mathog


  2. Re: imaging computer generates bogus ifcfg-eth1

    On Tue, 13 Feb 2007 13:21:19 -0800, David Mathog wrote:
    > Mandriva 2007. harddrake and haldaemon are turned off. I made an image
    > of one computer onto another, and after the image was written to disk
    > the script creating that image changed only one file:
    >
    > /etc/sysconfig/network
    >
    > which gets a new HOSTNAME value but is itself otherwise unchanged.
    >
    > Then the system reboots. When it comes up it now has an
    >
    > /etc/sysconfig/network-scripts/ifcfg-eth1
    >
    > entry (set for dhcpd).
    >
    > Why??? There's only one NIC on this system. Which part of
    > Mandriva 2007 took it upon itself to create this bogus ifcfg entry?


    Well, if you were to take a peek at all the Info discriptions in the
    Mandriva Control Center-->Services you might find a service which
    scans for new hardware. I would guess it found an on-board device and
    decided it was eth1.


  3. Re: imaging computer generates bogus ifcfg-eth1

    Bit Twister wrote:

    >
    > Well, if you were to take a peek at all the Info discriptions in the
    > Mandriva Control Center-->Services you might find a service which
    > scans for new hardware. I would guess it found an on-board device and
    > decided it was eth1.


    Isn't that "harddrake"? That service was turned off with chkconfig
    on the node that was imaged, and so in theory it should never have
    run on the clone. I don't recall seeing it turn on.

    The only thing that makes sense to me is that somewhere, somehow there's
    a date/time stamp test on
    /etc/sysconfig/network
    compared to
    /etc/sysconfig/network-scripts/ifcg*,
    and if the former is newer than the latter harddrake, or something like
    it, runs.

    Thanks,

    David Mathog

  4. Re: imaging computer generates bogus ifcfg-eth1

    David Mathog wrote:

    Imaged another 19 nodes, including reimaging the node that caused the
    glitch last time. This time after cleaning out various temporary files
    under /var in the master image (/var/lock/subsys/*, for instance).
    Mind you, aside from the NIC MAC addresses these machines are identical.
    18 of the nineteen came up without a problem, including the one which
    was a problem the first time it was imaged. The 19th one had the exact
    same eth0/eth1 confusion as was observed with the creation of a new
    ifcfg-eth1 file. I looked a bit more closely at the running OS after
    this glitch, and not only had it created an eth1 device, it was actually
    USING that device and it was working! Somehow or other it managed to
    move the single NIC from eth0 to eth1.

    To clean it up all that was required was:

    0. reboot
    1. boot failsafe
    2. rm -f /etc/sysconfig/network-scripts/ifcfg-eth1
    3. exit

    and then the network came up fine on eth0, as it should have in the
    first place.

    Pretty bizarre OS behavior.

    Regards,

    David Mathog


  5. Re: imaging computer generates bogus ifcfg-eth1

    David Mathog wrote:

    > David Mathog wrote:
    >
    > Imaged another 19 nodes, including reimaging the node that caused the
    > glitch last time. This time after cleaning out various temporary files
    > under /var in the master image (/var/lock/subsys/*, for instance).
    > Mind you, aside from the NIC MAC addresses these machines are identical.
    > 18 of the nineteen came up without a problem, including the one which
    > was a problem the first time it was imaged. The 19th one had the exact
    > same eth0/eth1 confusion as was observed with the creation of a new
    > ifcfg-eth1 file. I looked a bit more closely at the running OS after
    > this glitch, and not only had it created an eth1 device, it was actually
    > USING that device and it was working! Somehow or other it managed to
    > move the single NIC from eth0 to eth1.
    >
    > To clean it up all that was required was:
    >
    > 0. reboot
    > 1. boot failsafe
    > 2. rm -f /etc/sysconfig/network-scripts/ifcfg-eth1
    > 3. exit
    >
    > and then the network came up fine on eth0, as it should have in the
    > first place.
    >
    > Pretty bizarre OS behavior.


    Not really. Somewhere (I just didn't find out, but had it once) it's storing
    the MAC and on network-startup creates a new interface instead of
    overwriting, if it finds a different one.
    --
    vista policy violation: Microsoft optical mouse found penguin patterns
    on mousepad. Partition scan in progress to remove offending
    incompatible products. Reactivate MS software.
    Linux 2.6.17-mm1,Xorg7.1/nvidia [LinuxCounter#295241,ICQ#4918962]

+ Reply to Thread