Hi, after a lot of head scratching I found that the ifup-eth script in the initscripts package seems to be corrupt. I filed a bug with the details here: http://bugs.centos.org/view.php?id=5333
Given the nature of the corruption you might be not affected at all, see strange error messages or end up with networking problems on Centos 5.7 depending on what you configuration looks like.
I just thought I'd mention it here in case someone is in the same position as me and sees weird things happening when dealing with network configuration. I almost questioned my sanity before I went and dug into the networking scripts and found the problem.
Regards, Dennis
What kind of weird things?
I just finally got several boxes upgraded from 5.3 to 5.7 and so far have not seen anything odd.
On 12/21/2011 08:10 PM, Alan McKay wrote:
What kind of weird things?
I just finally got several boxes upgraded from 5.3 to 5.7 and so far have not seen anything odd.
I have the following config for a vlan interface: DEVICE=eth0.6 BOOTPROTO=none ONBOOT=yes TYPE=Ethernet VLAN=yes BRIDGE=vlanbr6
and a bridge: DEVICE=vlanbr6 BOOTPROTO=none ONBOOT=yes TYPE=Bridge STP=off NM_CONTROLLED=no
On Centos 5.6 this works as expected. On Centos 5.7 this also works but also outputs "RTNETLINK answers: No such device".
It seems the script is not broken but the code path executed is now very different (due to the missing "exec" call in the script). As a result the script at the end does a "ip route replace" on the bridge which apparently doesn't work. In the old version this code also exists but never gets run because the execution of "ifup-eth" ends with the "exec" line.
Regards, Dennis
On Wed, Dec 21, 2011 at 2:07 PM, Dennis Jacobfeuerborn dennisml@conversis.de wrote:
On Centos 5.6 this works as expected. On Centos 5.7 this also works but also outputs "RTNETLINK answers: No such device".
It seems the script is not broken but the code path executed is now very different (due to the missing "exec" call in the script). As a result the script at the end does a "ip route replace" on the bridge which apparently doesn't work. In the old version this code also exists but never gets run because the execution of "ifup-eth" ends with the "exec" line.
The upstream bugzilla that made the change you referenced is this one:
https://bugzilla.redhat.com/show_bug.cgi?id=684909
and the actual patch is:
http://git.fedorahosted.org/git/?p=initscripts.git;a=commitdiff;h=eb389d68f4...
I think you should file a bug report upstream because CentOS makes modifications only by inheriting the upstream (RH) code.
Akemi
-
Akemi Yagi -
Alan McKay -
Dennis Jacobfeuerborn