Random ending up with ifcfg-eth*.bak files...

Spike_White at Dell.com Spike_White at Dell.com
Wed Jan 25 16:14:34 CST 2012

Something Narendra just said reminded me of a "nuisance-level" bug I'm seeing on beaucoup of our RHEL5 Linux servers.

Randomly we end up w/ ifcfg-eth*.bak files in our /etc/sysconfig/network-scripts dir.  Been bitten by ifup trying to plumb up ifcfg-eth0.bak files before (in older RHEL versions).  So we *religiously* move old copies of ifcfg-eth* to a subdir or other save dir.  Not this dir. So it's not system engineers doing this.    

(BTW, I *think* that RHEL 5.7 has fixed this, that it no longer attempts to bring up ifcfg-eth0.bak interfaces.)

But these *.bak files are a nuisance.  Appears to be something internal to the network helper scripts that are creating these.  But I'm not seeing it, nor can I identify the trigger.  I thought the trigger was when we flip from DHCP to static, but it doesn't appear to be that.

These random ifcfg-eth*.bak files appear to have started around RHEL 5.5 or 5.7, didn't see it before that.

Here's an example lab server:
   [root at austl10geqldb02 network-scripts]# ls -l ifcfg-*
   -rw-r--r-- 1 root root 112 Sep 27 15:17 ifcfg-bond0
   -rw-r--r-- 1 root root 181 Sep 28 14:31 ifcfg-eth0
   -rw-r--r-- 1 root root 127 Oct 26 09:56 ifcfg-eth0.bak
  -rw-r--r-- 1 root root 145 Sep 27 15:12 ifcfg-eth1
  -rw------- 1 root root 119 Oct 26 09:56 ifcfg-eth1.bak
  -rw-r--r-- 1 root root 158 Sep 27 15:15 ifcfg-eth3
  -rw-r--r-- 1 root root  97 Oct 26 09:56 ifcfg-eth3.bak
  -rw-r--r-- 1 root root 158 Sep 27 15:15 ifcfg-eth5
  -rw-r--r-- 1 root root  97 Oct 26 09:56 ifcfg-eth5.bak
  -rw-r--r-- 1 root root 178 Oct 26 10:17 ifcfg-iscsi0
  -rw-r--r-- 1 root root 178 Oct 26 10:17 ifcfg-iscsi1
  -rw-r--r-- 1 root root 254 Jun  7  2011 ifcfg-lo

Here's ifcfg-eth0 and ifcfg-eth0.bak.  Differ only by IPADDR and HWADDR:

   [root at austl10geqldb02 network-scripts]# cat ifcfg-eth0
   # Broadcom Corporation NetXtreme II BCM5708S Gigabit Ethernet
   [root at austl10geqldb02 network-scripts]# cat ifcfg-eth0.bak
   [root at austl10geqldb02 network-scripts]#

We have flex-addressing enabled on our blades, so when these blade servers shift slots their MAC addresses change.  So that's why the above MAC address changed (for this blade server).  But I've seen these *.bak files also on RHEL 5.7 rack-able servers.  So I don't think the trigger is just an MAC address change.


More information about the Linux-PowerEdge mailing list