[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#607551: [linux-image-2.6.32-5-amd64] IPV6 and ethernet bonding



On Sun, 2010-12-19 at 18:48 +0100, Sylvain Avril wrote:
> Package: linux-image-2.6.32-5-amd64 
> Severity: normal 
> 
> --- Please enter the report below this line. --- 
> 
> Hello,
> 
> I tried to setup IPv6 on a server which use ethernet bonding.
> 
> My host use eth0 and eth1 which combine into device bond0.
> When I try to add an IPv6 address to bond0 (ip -6 addr add
> xxxxxxxxxxxxxxxxxxxxxxxxx/64 dev bond0), dmesg show this error:
> [1492995.485604] bond0: IPv6 duplicate address
> xxxxxxxxxxxxxxxxxxxxxxxxx detected!
> This address cannot be ping from this host or another host.
> 
> If I remove eth1 from bonding (ifenslave -d bond0 eth1), delete and
> re-add this address, everything work as expected.
> 
> After some research, it seem that it's a bug corrected in 2.6.33.
> 
> You can find more informations in a similar bug report for Redhat :
> https://bugzilla.redhat.com/show_bug.cgi?id=516985
> You will find in it the relevant patches for the kernel.

We normally prefer to apply bug fixes that have already been applied
upstream.  It looks like the relevant changes are:

commit e9d3e084975869754d16f639378675c353560be9
Author: Herbert Xu <herbert@gondor.apana.org.au>
Date:   Tue May 18 15:36:06 2010 -0700

    ipv6: Replace inet6_ifaddr->dead with state

commit 4c5ff6a6fe794f102479db998c69054319279e3c
Author: Herbert Xu <herbert@gondor.apana.org.au>
Date:   Tue May 18 15:54:18 2010 -0700

    ipv6: Use state_lock to protect ifa state

commit f2344a131bccdbfc5338e17fa71a807dee7944fa
Author: Herbert Xu <herbert@gondor.apana.org.au>
Date:   Tue May 18 15:55:27 2010 -0700

    ipv6: Use POSTDAD state

commit 622ccdf107bcb49c4d8fb65512652566d4c8928a
Author: Herbert Xu <herbert@gondor.apana.org.au>
Date:   Tue May 18 15:56:06 2010 -0700

    ipv6: Never schedule DAD timer on dead address

We would also add:

commit 853dc2e03ddd0c49885ed55c48755d2b1073122a
Author: Ursula Braun <ursula.braun@de.ibm.com>
Date:   Sun Oct 24 23:06:43 2010 +0000

    ipv6: fix refcnt problem related to POSTDAD state

However, we'll have to decide whether this is serious enough to fix for
'squeeze'.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: