My eth0:1 IP taking precedence for outgoing connections over eth0

Ex. I have 2 IPs on the server, both assigned to eth0

when I ssh to another server, and hit w, it shows I am connected as the IP of eth0:1

Also when sending mail, it has the IP or eth0:1, instead of the servers main IP of eth0, which is causing mail to bounce for no PTR record.

I know I can add a PTR to the eth0:1 IP, but do you know how I can get hte IP to show as the eth0 IP for all outbound connections?

I stopped the firewall and even ran an iptables -F but the same thing still happens.

TIA!

Can you show ifconfig and route outputs

Sure:
# ifconfig
eth0 Link encap:Ethernet HWaddr 00:0F:EA:F4:F3:40
inet addr:xxx.xx.7.106 Bcast:xxx.xx.7.255 Mask:255.255.255.0
inet6 addr: fe80::20f:eaff:fef4:f340/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:55819384 errors:0 dropped:0 overruns:499 frame:499
TX packets:11694775 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:230216797 (219.5 MiB) TX bytes:2373278725 (2.2 GiB)

eth0:1 Link encap:Ethernet HWaddr 00:0F:EA:F4:F3:40
inet addr:xxx.xx.6.154 Bcast:xxx.xx.6.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1

There are actually 21 IPs on that server, I have truncated ifconfig, but the output from route is kinda long:
# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
xxx.xx.7.222 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.107 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.6.147 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.108 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.14 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.109 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.168 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.110 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.111 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.112 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.6.174 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.167 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.113 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.114 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.115 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.6.173 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.6.155 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.6.154 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.7.211 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.6.184 * 255.255.255.255 UH 0 0 0 eth0
xxx.xx.6.0 * 255.255.255.0 U 0 0 0 eth0
xxx.xx.7.0 * 255.255.255.0 U 0 0 0 eth0
169.254.0.0 * 255.255.0.0 U 0 0 0 eth0
default xxx.xx.7.1 0.0.0.0 UG 0 0 0 eth0

Actually what is happening is when it connects to a server with the IP on the same cblock as eth0, it uses the servers main IP, if it connects to a server on the same cblock as eth0:1, it uses the IP of eth0:1

If it connects to an IP not on the cblock of the eth0 or eth0:1, it uses the IP of eth0.

Is there anyway to get it to use eth0 all the time?

Can't you delete the route for eth0:1?

I suspect that given two equivalent routes, the kernel will use the last one added to the table.

You should be setting the netmask of the aliases to 255.255.255.255 if you don't want it to use those IPs for outbound connections.

Thanks, I'll try that.