Thanks Pedro,

I'll ask Babu to take a look tomorrow.

Foucault
Le 15 juil. 2015 19:49, "Pedro Marques" <[email protected]> a écrit :

>  Foucault,
> For the existing LBaaS v1 API, it seems to me that the best option is the
> one you propose:
>   - Automatically add a redirect when the protocol is HTTPS so that 80
> redirects to 443 (and potentially 8080 to 8143).
>   - We should have this behavior be determined by a configuration option
> in contrail-vrouter-agent.conf.
>
>  The LBaaS v2 API introduces the concept of a listener object (i.e. a
> front-end port) such that there can be multiple ports per load-balancer
> object. Once we add support for the v2 API this can be controlled
> explicitly.
>
>     Pedro.
>
>   On Jul 15, 2015, at 10:21 AM, Foucault de Bonneval <
> [email protected]> wrote:
>
>  Hi all,
>
>  I have a use case that seems difficult to address with actual LBaaS
> implementation.
> First of all, this is a limitation from Neutron and LBaaS implementation
> is based on it.
>
>  My use case is this one :
> * I would like my web users to always get in HTTPS
> * my website is at http(s)://www.me.com
>
>  But when I want to create my LB, i create a
>  * lb-pool
>  * lb-vip
>  * members
> lb-pool and lb-vip can manage only one of --protocol {TCP,HTTP,HTTPS}
>
>  I cannot get the same VIP (an so floating-ip) to listen both tcp/80 and
> tcp/443
>
>  So I won't be able to redirect from http://www.me.com/ =>
> https://www.me.com/
>
>  Does anyone have a native workaround about this issue ?
>
>
>
>  I was thinking, as Openstack community is not talking a lot about this
> point that we could add a case in Opencontrail vrouter/config/lbaas stack :
> - when a user asks for a tcp/443 lb-pool, lb-vip
> - automatically provision a tcp/80 passthrough in HAproxy
>
>  Based on the assumption that someone who does tcp/443 passthrough have a
> strong intention to do 80 -> 443 redirection.
>
>  What do you think of this (dirty) solution ?
> I can file a launchpad bug and get someone of my team to work on it if
> some of you agree on it.
>
>  Thanks,
> Foucault
>  _______________________________________________
> Dev mailing list
> [email protected]
> http://lists.opencontrail.org/mailman/listinfo/dev_lists.opencontrail.org
>
>
>
_______________________________________________
Dev mailing list
[email protected]
http://lists.opencontrail.org/mailman/listinfo/dev_lists.opencontrail.org

Reply via email to