Hello all,

We have a Linux box running Centos with 3 network interfaces with static IP assigned to each,

eth0 192.168.0.25
eth1 172.17.1.10
eth2 172.19.1.10

whenever we reboot the box, it seem that the IPs between the eth0 and eth1 get swapped, it would become

eth0 172.17.1.10
eth1 192.168.0.25
eth2 172.19.1.10

one the sys admins at work said that depending on which interface comes up first gets assigned the first ip, but that just doesn't sound right to me. I suspect something is not configured properly. What could be the problem? Where should I look to fix this?

Thanks very much in advance!

I think he's on the right track, Check the persistent net rules.. /etc/udev/rules.d/70-persistent-net.rules

http://reactivated.net/writing_udev_rules.html

Given that your ethernet device names aren't changing, I'm having trouble understanding how it could be a udev problem. IOW, if your networking script (however that works on Centos) says that eth0 should be 192.168.0.25, then that's what it should be, regardless of what order eth0 comes up in. That's not to say that device 1 (which you want to be eth0) couldn't be the wrong device, and thus get the wrong IP as a result of udev. Or is there something you're not telling us to avoid some confusion?

It might be more helpful if we could see the results of /sbin/ifconfig on two occasions when they're different.

comparing the IP address and MAC address for each interface when addressed correctly, and incorrectly, would make it obvious if the interface id's are swapping.

As quakeboy02 said it's hard to tell from here without the proper info though. so I'll consult my Magic Eight Ball in lou of factual data. I asked the Magic Eight Ball if the interfaces were swapping position at boot and it responded: "Definitely !" The Magic Eight Ball is rarely ever wrong... so unless we get more info, that is what we will have to go by.





Aw come on you have to admit that was humorous... Well, at least I thought it was..

Wait till I get over the flu to do that, will ya?!

Cubehead are you out there ?

Were you able to check the interface info to get a better idea of what is happening ? ....
Did you already resolve the issue ?
enquiring minds want to know


Quakeboy, you didn't get the H1N1 (swine flu) did you ? Hope you're feeling better soon..

I haven't used Centos (Red Hat derivative, yes?) but I think that if you look in the ifcfg-ethx files (where x is 0 1 or 2 for you) you have the option to add the line:See more here: http://www.redhat.com/docs/manuals/e...nterfaces.html

This looks easier than messing with udev