eth0 alias issue

[T. Howell-Cintron]

Greetings fedora-list!

I'm trying to add another IP address to one of my machines running FC11
and have run in to a difficult problem.  I'm used to simply copying a
working ifcfg-eth0 to ifcfg-eth0:1, changing DEVICE, NAME, and IPADDR
and it works as expected.  Not this time around..

/etc/sysconfig/network-scripts/ifcfg-eth0:
DEVICE=eth0
BOOTPROTO=none
DNS1=69.58.0.4
GATEWAY=69.58.21.129
HWADDR=00:10:4B:95:A1:99
IPADDR=69.58.21.130
NETMASK=255.255.255.240
ONBOOT=yes
ETHTOOL_OPTS="speed 100 duplex full autoneg off"
TYPE=Ethernet

/etc/sysconfig/network-scripts/ifcfg-eth0:1
DEVICE=eth0:1
NAME=eth0:1
IPADDR=69.58.21.135
NETMASK=255.255.255.240
ONBOOT=yes
TYPE=Ethernet

After editing the file I run "ifup eth0:1" and it's bound as expected.
Running "ip addr" confirms this..

2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UNKNOWN qlen 1000
    link/ether 00:10:4b:95:a1:99 brd ff:ff:ff:ff:ff:ff
    inet 69.58.21.130/28 brd 69.58.21.143 scope global eth0
    inet 69.58.21.135/28 brd 69.58.21.143 scope global secondary eth0:1
    inet6 fe80::210:4bff:fe95:a199/64 scope link
       valid_lft forever preferred_lft forever

But when I reboot things go wonky.  I can only access it via the new
aliased IP address, and when I do the output of "ip addr" shows it's
bound to eth0 and there is no mention of any aliases.

2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UNKNOWN qlen 1000
    link/ether 00:10:4b:95:a1:99 brd ff:ff:ff:ff:ff:ff
    inet 69.58.21.135/28 brd 69.58.21.143 scope global eth0
    inet6 fe80::210:4bff:fe95:a199/64 scope link
       valid_lft forever preferred_lft forever

So I run "ifup eth0" and it responds with..
SIOCGIFADDR: Cannot assign requested address
SIOCSIFBROADCAST: Cannot assign requested address
SIOCSIFBRDADDR: Cannot assign requested address
SIOCSIFFLAGS: Cannot assign requested address

.. and now "ip addr" reports:

2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UNKNOWN qlen 1000
    link/ether 00:10:4b:95:a1:99 brd ff:ff:ff:ff:ff:ff
    inet 69.58.21.135/28 brd 69.58.21.143 scope global eth0
    inet 69.58.21.130/28 brd 69.58.21.143 scope global secondary eth0
    inet6 fe80::210:4bff:fe95:a199/64 scope link
       valid_lft forever preferred_lft forever

.. so I try "service network restart" to see what happens.  Things stay
the same - both IP addresses are bound to eth0 and are working as
expected, but still no mention of eth0:1.  It gets even stranger as when
I "mv ifcfg-eth0:1 tmp-eth0" within a few seconds the second IP address
stops responding, as if some daemon is watching that file and drops the
interface as soon as it's removed.  The output of "ip addr" confirms this:

2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UNKNOWN qlen 1000
    link/ether 00:10:4b:95:a1:99 brd ff:ff:ff:ff:ff:ff
    inet 69.58.21.130/28 brd 69.58.21.143 scope global eth0
    inet6 fe80::210:4bff:fe95:a199/64 scope link
       valid_lft forever preferred_lft forever

I'm at a complete loss as to why it's behaving this way.  Any ideas?

Thanks,
Tom Howell-Cintron