The reproducer doesn't seem to be sufficient to reproduce the issue. Also, iproute just queries the kernel and displays the answer.

Just a note. I've seen similar device names when checking out 'ip address save' and 'ip address restore'. It looks like a device naming scheme internal to iproute2 that is sometimes used instead of kernel based interface name translation, probably to avoid translation when it might not make sense.

Sorry, I have no idea, try jaster, who tests several last errata for iproute.

Hi,

I tried to reproduce it, but I failed.

Hi Konstantin,

(In reply to Konstantin Volkov from comment #0)
> Description of problem:
> Sometimes iproute reports wrong device name
> 
> Version-Release number of selected component (if applicable):
> iproute-2.6.32-31.el6
> 
> How reproducible:
> ~1 of 30000

I tried to reproduce this in my local RHEL6 VM with iproute-2.6.32-45.el6 installed using the following command:

# for i in {1..100000}; do ip route list table all | grep if; done

It didn't happen a single time for me. Can you still reproduce the issue with a current RHEL6 version of iproute? Maybe it has been fixed as a side-effect of some other fix.

Hi Phil!

Well, we stop catch this issue with downgrade to iproute-2.6.32-23.el6 . I'll ask QA to restart test with latest iproute-2.6.32-45.el6 . 

PS Looks like ubuntu guys already fixed this problem, look at https://bugs.launchpad.net/ubuntu/+source/iproute/+bug/1281366 , there is also link to some test script.

Good luck,
Konstantin

Hi Konstantin,

OK, so I tried to reproduce the issue after downgrading iproute to version 2.6.32-31.el6, also unsuccessful. To make sure I put a decent amount of load into the test, I also tried creating 1000 dummy interfaces, each with an IPv4 address assigned (in addition to the automatic link-local IPv6 ones). Consequently, 'ip route show table all' emits over 4000 lines of output.

Maybe this really is a kernel issue which is fixed meanwhile. Since I don't know which kernel your faulty system runs, I tested using kernel-2.6.32-573.7.1.el6.x86_64.

In order to advance this, could you please provide more information about the failing system? Especially kernel version, number and types of interfaces and ip address and route configuration are interesting. General environment might have an influence as well, like general system load, etc.

Of course I am eager to hearing what results came out when testing with iproute-2.6.32-45.el6. When do you expect them to be available?

Thanks, Phil

Hi Phil!

Looks like i can't provide you exact version of kernel and other environment parameters of the test failed ~1.5 years ago... I just asked our QA, they confirm that on the latest iproute no similar failures found.

So, you can close it as can't reproduce.

Good luck,
Konstantin

Hi Konstantin,

(In reply to Konstantin Volkov from comment #13)
> Looks like i can't provide you exact version of kernel and other environment
> parameters of the test failed ~1.5 years ago... I just asked our QA, they
> confirm that on the latest iproute no similar failures found.

Many thanks for looking into this again, even after such a long time.

> So, you can close it as can't reproduce.

Good to know the problem is gone (for whatever reason).

Thanks, Phil