[Users] Legacy OpenVZ CTs lose IPv6 when restarted/created

[Peter Laws]

Here's a tcpdump from host while I was using ping6 from a CT that has
IPv6 outbound versus a CT that _cant_ connect IPv6 outbound

https://pastebin.com/AiqU8N6k

As you can see, I'm not getting any "echo reply"'s from the CT that
can't connect outbound.

Regards,

Peter Laws


On 4 December 2017 at 07:53, Vasily Averin <vvs at virtuozzo.com> wrote:
> Dear Peter,
>
> seems something was changed, either on your node or on external routers.
> We had no updates on openVz userspace long time ago
> and I doubt it can be caused by kernel update.
> Therefore I believe it was caused by some changes in network infrastructure.
>
> I would advice you to run tcpdump in all possible key points,
> it allows to understand where exactly your packets were lost.
>
> Thank you,
>         Vasily Averin
>
> On 2017-12-02 13:28, Peter Laws wrote:
>> Hello,
>>
>> Before I go bald, I would like some advice on why IPv6 in containers
>> _suddenly_ stopped working when restarting and creating new CTs.
>>
>> - I can ping a CT assigned IPv6 (from the outside and on HN), but I
>> cant ping/connect out (but can ping HN);
>> - restarting or creating a CT results in the above scenario;
>> - HN ip6tables -L is essentially blank, and ip6tables -L (output) in
>> CT is set to allow;
>> current running CTs are fine.
>>
>> Anyone help?
>>
>> HN IPv6 settings look ok and have not been touched.
>>
>>
>> Note: OVH server.
>>
>>
>> Regards,
>>
>> Peter Laws
>> _______________________________________________
>> Users mailing list
>> Users at openvz.org
>> https://lists.openvz.org/mailman/listinfo/users
>>