Red Hat Bugzilla – Bug 314541
dangling symlink for veth2
Last modified: 2009-12-14 16:25:06 EST
Description of problem:
This link is broken:
/dev/.udev/failed/class@net@veth2 -> /sys/class/net/veth2
The link is not necessarily the main issue, since veth2 does not exist.
However, the system has veth0, veth1, veth3. I'm unsure why veth2 is missing.
I haven't noticed any functional issues with this setup, but the missing veth2
is somewhat worrisome.
Version-Release number of selected component (if applicable):
RHEL5.1 beta s8 (kernel 2.6.18-48.el5xen)
How reproducible: 1/1
Steps to Reproduce:
1. Install RHEL5.1 beta s8, with Xen / install key, default settings.
2. (unsure if intermediate steps are applicable)
3. Check for dangling symlinks in /dev/.udev/
Actual results: veth2 symlink points at nonexistent /sys entry
Expected results: /sys entry should exist, but in case it does not, the symlink
should not exist either.
I created a paravirtual guest before I noticed the missing veth2. I also
replaced "-- rhgb quiet" with "dom0_mem=4G -- debug" in elilo.conf. I'm not yet
sure if either of these are relevant.
Created attachment 212701 [details]
sysreport -a (.tar.bz2)
I encountered this today using release candidate 1. The missing veth device was
veth0, but everything else appeared to be the same.
FWIW, this is still present in RHEL5.2 beta 3. After a default install with
Xen, I ended up with no veth0.
It doesn't appear to have any functional impact, though. At least, not that
I've noticed. Is it intended behavior?
This is an artifact of the way Xen's networking works.
When you initially boot there is just the real physical device 'eth0'
Then Xen runs and loads its netloop module. This creates veth0 -> veth3 and
vif0.0 -> vif0.3
Then Xen's network-bridge script runs. This renames 'eth0' to 'peth0'. Adds
'peth0' and 'vif0.0' to a bridge 'xenbr0'. The 'veth0' is then renamed to 'eth0'.
This last step causes the dangling symlink and is harmless