Cloud instance with reduced MTU, container uses 1500, connections stall

[Andreas Hasenack]

On Mon, Jun 16, 2014 at 9:40 AM, Andreas Hasenack <andreas at canonical.com>
wrote:

>
>
>
> On Fri, Jun 13, 2014 at 5:01 PM, Andreas Hasenack <andreas at canonical.com>
> wrote:
>
>> Hi,
>>
>> I'm not sure where or how this should be fixed.
>>
>> I have a case where openstack was juju deployed with neutron networking,
>> and we needed to set instance-mtu = 1454 in the neutron-gateway charm.
>>
>>
> For reference:
> http://techbackground.blogspot.co.uk/2013/06/path-mtu-discovery-and-gre.html
>
>
The right ICMP packet is generated, but that's it.

This is what I see on neutron-gateway:

19:13:05.600151 IP 10.96.1.4 > 10.96.1.4: ICMP 10.96.1.9 unreachable - need
to frag (mtu 1500), length 556

These are the actors:
10.96.0.10: MAAS server (10.96.0.0/21)
10.96.1.4: neutron-gateway's br0 (MTU 1500)
10.96.1.9: nova-compute where my instance is running (br0, MTU also 1500)
10.10.0.6 (not shown): my instance, running on 10.96.1.9 (eth0, MTU 1454).
wget http://10.96.0.10/ works
10.0.3.251 (not shown): lxc on instance above, MTU 1500, running wget
http://10.96.0.10/. It connects (tcp handshake) but gets stuck.

I don't see that ICMP packet anywhere else.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-server/attachments/20140616/d45bf0d6/attachment.html>