I think a recent update changed the behavior of ifup on systems where the drives were moved or cloned and at boot time have a different MAC address than the one found in the corresponding interface file under /etc/sysconfig/network-scripts or /etc/sysconfig/networking/profiles/default/. Now the interface is ignored with an error message about having a different MAC address than expected. I frequently update remote machines by shipping configured swappable drives so this becomes an interesting problem.
However, it seems to be intended behavior so instead of reporting it as a bug, I'd like to know the correct procedure to prepare a drive to run in an unknown target machine and where this might be documented.
I had a similar behaviour when I copied the ifcfg-eth... files from a FC1 installation on the same machine after CentOS 4 installation. I suspect they changed the sequence of scanning the pci bus.I mean from bottom up and vice versa. Fortunately I had console access.
On Sat, 2005-04-23 at 09:18 +0300, Nikos Zaharioudakis wrote:
I had a similar behaviour when I copied the ifcfg-eth... files from a FC1 installation on the same machine after CentOS 4 installation. I suspect they changed the sequence of scanning the pci bus.I mean from bottom up and vice versa. Fortunately I had console access.
This should only happen if you supply a value for HWADDR in your ifcfg- eth<n> config file. You either need to make sure it matches the new hardware, or take it out.
The point of this code is for the case where some system change causes an existing interface to be detected in a different order than some earlier case, or someone simply wants to specify what name a specific piece of hardware gets. If you only have one ethernet adapter, there's no need to specify its MAC address, unless you want to name it something other than eth0. And if you have multiple adapters and you are happy with the names assigned during installation, the names will be consistent except under a few specific situations.
On Sat, 2005-04-23 at 23:14, C. Linus Hicks wrote:
On Sat, 2005-04-23 at 09:18 +0300, Nikos Zaharioudakis wrote:
I had a similar behaviour when I copied the ifcfg-eth... files from a FC1 installation on the same machine after CentOS 4 installation. I suspect they changed the sequence of scanning the pci bus.I mean from bottom up and vice versa. Fortunately I had console access.
This should only happen if you supply a value for HWADDR in your ifcfg- eth<n> config file. You either need to make sure it matches the new hardware, or take it out.
I didn't set it on purpose anywhere, but it ended up in /etc/sysconfig/networking/profiles/default/ifcfg-eth0, apparently from the initial setup. After the drive was copied, the IP address was set with /usr/sbin/redhat-config-network-gui which until a recent update was all that was necessary.
The point of this code is for the case where some system change causes an existing interface to be detected in a different order than some earlier case, or someone simply wants to specify what name a specific piece of hardware gets. If you only have one ethernet adapter, there's no need to specify its MAC address, unless you want to name it something other than eth0. And if you have multiple adapters and you are happy with the names assigned during installation, the names will be consistent except under a few specific situations.
The machines in question have multiple NICs but only one is used - or it was until the update decided to ignore it...
-
C. Linus Hicks -
Les Mikesell -
Nikos Zaharioudakis