"ip -6 addr flush" flushes much more than just the addresses

Bug #1023174 reported by Stéphane Graber on 2012-07-10
14
This bug affects 3 people
Affects Status Importance Assigned to Milestone
linux (Ubuntu)
Medium
Unassigned

Bug Description

root@testv6-tester:~# ls /proc/sys/net/ipv*/conf/
/proc/sys/net/ipv4/conf/:
all default eth0 lo sit0

/proc/sys/net/ipv6/conf/:
all default eth0 lo sit0
root@testv6-tester:~# ip addr flush dev eth0
root@testv6-tester:~# ls /proc/sys/net/ipv*/conf/
/proc/sys/net/ipv4/conf/:
all default eth0 lo sit0

/proc/sys/net/ipv6/conf/:
all default lo sit0

This results in no IPv6 connectivity on the interface and no obvious way to get it back to working condition other than destroying and recreating the interface (usually by rebooting).

I reproduced this on the stable 12.04 kernel (3.2.0-26-generic) as well as 12.10 (3.5.0-3-generic).

This is currently breaking ipv6 connectivity for some dual-stack setups in debian-installer though only for Ubuntu 12.10 (because the flush call was ipv4-specific in 12.04). I'll workaround the issue in d-i, but this bug should really be looked into.

Related branches

Stéphane Graber (stgraber) wrote :

Adding bot-stop-nagging as I don't feel like being spammed by Brad's bot, feel free to ping on IRC.

I reproduced this on a variety of systems, all of them with IPv6 connectivity. This happens on virtual interfaces (veth, bridge, bond, ...) just as much as on physical interface (eth0 on my netbook).

tags: added: bot-stop-nagging

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 1023174

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

Changed in linux (Ubuntu):
status: New → Incomplete
tags: added: precise
Stéphane Graber (stgraber) wrote :

Hmm, that bot ignored bot-stop-nagging...

Anyway, it's a good question regardless, dmesg didn't contain any error message or anything related to network interfaces.

Changed in linux (Ubuntu):
status: Incomplete → Confirmed
Stéphane Graber (stgraber) wrote :

Uploaded a new isc-dhcp working around the issue in the debian-installer case. A debian-installer upload will be needed to get this workaround included.

dhclient will still try to run "ip -6 addr flush" in some DHCPv6 cases so this is still a important to get this fixed soon before IPv6 becomes more popular and people start noticing their connectivity vanishing when dhclient exits.

tags: added: quantal rls-q-incoming
Joseph Salisbury (jsalisbury) wrote :

Can you also give the latest upstream kernel a test, which can be downloaded from:
http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.5-rc6-quantal/

Changed in linux (Ubuntu):
importance: Undecided → Medium
tags: added: kernel-key

Same thing with the mainline build.

root@lantea:~# ls /proc/sys/net/ipv6/conf/
all bond0 bond0.1005 br1005 default eth0 eth1 lo

root@lantea:~# ip addr flush dev br1005

root@lantea:~# ls /proc/sys/net/ipv6/conf/
all bond0 bond0.1005 default eth0 eth1 lo

root@lantea:~# uname -a
Linux lantea 3.5.0-030500rc6-generic #201207072135 SMP Sun Jul 8
01:43:16 UTC 2012 i686 i686 i686 GNU/Linux

Joseph Salisbury (jsalisbury) wrote :

If you have a chance, can you also give the v3.6-rc3[0] kernel a test?

[0] http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.6-rc3-quantal/

tags: removed: kernel-key
tags: added: kernel-bug-exists-upstream
Changed in linux (Ubuntu):
status: Confirmed → Triaged
Joseph Salisbury (jsalisbury) wrote :

The v3.6 kernel[0] has now been released. Can you test that kernel to see if the bug exists there?

Also, I will remove the rls-q-incoming tag for now, since the bug is upstream.

[0] http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.6-quantal/

tags: removed: rls-q-incoming
Pali (pali) wrote :

Tested on 3.12.0-5-generic kernel and problem is still there.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers