Centos server keeps jumping to address instead of static assigned address

Question

I have a headless Centos server that I've configured with a static IP address. I've disabled Network Manager, avahi-daemon etc. Whenever I restart the network it just jumps to 10.155.90.100 instead of the required 10.155.90.61 with no explanation:

Here is my ifcfg-eth0:

   # Intel Corporation 82545EM Gigabit Ethernet Controller (Copper)
    DEVICE=eth0
    BOOTPROTO=static
    ONBOOT=yes
    #HWADDR=00:50:56:b0:5a:57
    IPADDR=10.155.90.61
    NETMASK=255.255.255.0
    GATEWAY=10.155.90.1
    TYPE=Ethernet
    USERCTL=no
    IPV6INIT=no
    PEERDNS=yes
    NM_CONTROLLED=no

It seems to be running DHCP anyway. I cant seem to get a more detailed log than this in /var/log/messges:

Jul 11 12:50:33 NET[4767]: /sbin/dhclient-script : updated /etc/resolv.conf
Jul 11 12:50:34 kernel: 8021q: adding VLAN 0 to HW filter on device eth0
Jul 11 12:50:34 kernel: ADDRCONF(NETDEV_UP): eth0: link is not ready
Jul 11 12:50:34 kernel: e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
Jul 11 12:50:34 kernel: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
Jul 11 12:50:36 dhclient: DHCPREQUEST on eth0 to 255.255.255.255 port 67 (xid=0x39edae98)
Jul 11 12:50:36 dhclient: DHCPACK from 10.155.50.20 (xid=0x39edae98)
Jul 11 12:50:36 NET[5157]: /sbin/dhclient-script : updated /etc/resolv.conf
Jul 11 12:50:36 dhclient: bound to 10.155.90.100 -- renewal in 255205 seconds.

How can I turn off DHCP and stop the server jumping for the wrong IP?

Solution 1

plain and simple solution: deinstall what you dont need :)

yum remove dhclient 

Solution 2

You don't need PEERDNS since you're using a static IP. This is causing dhclient to run and update your resolv.conf file (as it's designed to do). In the process, it also requests an IP address when it requests the nameservers from DHCP.

Solution 3

To assign a fixed IP address, use none as the value for BOOTPROTO.

From the "RHEL5 Deployment Guide", section "Interface Configuration Files":

BOOTPROTO=protocol

where protocol is one of the following:

• none — No boot-time protocol should be used.
• bootp — The BOOTP protocol should be used.
• dhcp — The DHCP protocol should be used.

Solution 4

Maybe I'm a hack, but I like using the system-config-network-tui utility on RHEL/CentOS/RPM-based system for this, as it helps clean up any funky or hand-(mis)configured interface scripts.

View more solutions

Comments

[Dennis Nolte]

deinstall the service? dhclient

[blarg]

@DennisNolte. That fixed it!

[MadHatter]

I don't think that's right; I use BOOTPROTO=static very happily in all sorts of CentOS boxes, specifically both C5 and C6, with no problems.

[iSee]

Maybe he should have a look at the documentation i.e. less /usr/share/doc/initscripts-*/sysconfig.txt and search for BOOTPROTO. It says BOOTPROTO=none|bootp|dhcp on Scientific Linux 6.5.

[Nathan C]

It appears none or static is dependent on the OS. RHEL5 changed it, but from the research I've done (like this wiki) CentOS still uses static.

[ewwhite]

Or just stop the service.

[Andrew]

It will fix this problem but it's a bit "brute force". What if other interfaces need DHCP? I would investigate Nathan C's answer below about PEERDNS.