Hi,

Thanks guys for the tips. I can connect to haproxy with ECDSA cipher
using the following cipher string on the OpenSSL client side:

   openssl s_client -connect 127.0.0.1:10000 -cipher ECDSA:EECDH:ECDH

and this string in the haproxy configuration:

   ssl-default-bind-ciphers "ECDSA:EECDH:ECDH"

and only ECDSA certificates.

The negociated cipher is "AECDH-AES256-SHA", and I don't know if this
cipher is ECDSA :) At least it seems to work.

Thierry



On Tue, 11 Oct 2016 15:32:04 -0700
Bryan Talbot <bryan.tal...@playnext.com> wrote:

> 
> > On 12 Oct 2016 8:45 am, "Igor Cicimov" <ig...@encompasscorporation.com 
> > <mailto:ig...@encompasscorporation.com>> wrote:
> > >
> > > On 11 Oct 2016 7:05 pm, "Thierry Fournier" <thierry.fourn...@ozon.io 
> > > <mailto:thierry.fourn...@ozon.io>> wrote:
> > > > I'm currently trying to investigate about a little leak of memory in
> > > > the certificates loading, and I try to test ECDSA certificates and
> > > > cipher.
> > > >
> > > > I can't done this :( I don't understand anything in the ECDSA
> > > > certificate process.
> > > >
> > > > My test certificate is generated from a little chain where the root CA
> > > > is autosigned. So the root CA and the 2 intermediate are RSA
> > > > certificates. The ECDSA certificate is build with these commands:
> > > >
> > > >    openssl ecparam -name secp521r1 -genkey -param_enc explicit -out \
> > > >       $CN.ecdsa.key
> > 
> > 
> 
> 
> I ran into this as well and it turns out that s_client and s_server do not 
> seem to play nicely with curves when using -param_enc explicit and instead 
> prefer to only deal with named curves.
> 
> Encode the key params using named curve that both sides can accept and your 
> test should work.
> 
> Also, see 
> https://groups.google.com/forum/#!topic/mailing.openssl.users/Rg6yV4ccWeo 
> <https://groups.google.com/forum/#!topic/mailing.openssl.users/Rg6yV4ccWeo>
> 
> -Bryan
> 
> 
-- 
Thierry Fournier
Web Performance & Security Expert
m: +33 6 68 69 21 85      | e: thierry.fourn...@ozon.io
w: http://www.ozon.io/    | b: http://blog.ozon.io/

Reply via email to