Open Stack

Hello Jonathan,

Thanks for the tips. It is interesting that you point this out, it was indeed one of my concern. If I understood the process correctly, Certbot will run on each HAproxy node and request LetsEncypt to issue certificate on each node. This means that we ask many certificates for the same domain (for instance openstack.example.com <http://openstack.example.com/>). This must impact the following rate limit of LetsEncrypt:

"Renewals are treated specially: they don’t count against your Certificates per Registered Domain limit, but they are subject to a Duplicate Certificate limit of 5 per week. Exceeding the Duplicate Certificate limit is reported with the error message too many certificates already issued for exact set of domains. » LetsEncrypt Website

Does that mean that the the deployment is limited to 5 HAProxy nodes ? Normally we are safe tho, we have 3.

Concerning, the timeout values, we’ll make sure to check them out. We’ll upgrade to Wallaby or Xena by the end of the year in any case.

Thanks,
Marc-Antoine

> Le 22 févr. 2022 à 10:35, Jonathan Rosser <jonathan.rosser at rd.bbc.co.uk> a écrit :
> 
> Hi Marc-Antione,
> 
> No problem. I would recommend adding --staging to haproxy_ssl_letsencrypt_setup_extra_params whilst you get the letsencrypt support working. You will not get a proper certificate with that flag but it will bypass the letsencrypt rate limit so you can have as many tests as you need.
> 
> It would be also worth checking the timeout values on later branches, Ussuri is now in extended-maintenance so not receiving back ported bug fixes.
> 
> See for example https://github.com/openstack/openstack-ansible/blob/stable/xena/inventory/group_vars/haproxy/haproxy.yml#L248-L258 <https://github.com/openstack/openstack-ansible/blob/stable/xena/inventory/group_vars/haproxy/haproxy.yml#L248-L258>
> 
> On 21/02/2022 18:51, Marc-Antoine Godde wrote:
>> Thanks for your huge help. It’s is exactly what we wanted to try. We’ll feel more confident.
>> 
>> Best,
>> Marc-Antoine
>> 
>> 
>> 
>>> Le 21 févr. 2022 à 18:52, Jonathan Rosser <jonathan.rosser at rd.bbc.co.uk <mailto:jonathan.rosser at rd.bbc.co.uk>> a écrit :
>>> 
>>> Hi Marc-Antoine,
>>> 
>>> For setting the horizon acl, see https://docs.openstack.org/openstack-ansible/ussuri/user/security/index.html <https://docs.openstack.org/openstack-ansible/ussuri/user/security/index.html>
>>> Specifically:
>>> 
>>> "Copy the whole variable haproxy_default_services from /opt/openstack-ansible/inventory/group_vars/haproxy/haproxy.yml to /etc/openstack_deploy/group_vars/haproxy/haproxy_all.yml and update the section for horizon to include the ACL redirects http-01 challenges to the HAProxy letsencrypt backend as follows: ......"
>>> 
>>> It is correct that this is not necessary in later releases and the letsencrypt support is more straightforward to configure in Victoria.
>>> 
>>> You can also join #openstack-ansible IRC channel for some real-time help if needed.
>>> 
>>> Jonathan.
>>> 
>>> On 21/02/2022 17:25, Marc-Antoine Godde wrote:
>>>> Hello,
>>>> 
>>>> I have a question on how to setup LetsEncrypt with OpenStack Ansible. We are still on OpenStack Ussuri.
>>>> 
>>>> We added the following variables to user_variables.yml.
>>>> 
>>>> ==================================================================================
>>>> haproxy_ssl_letsencrypt_enable: True
>>>> haproxy_ssl_letsencrypt_install_method: "distro"
>>>> haproxy_ssl_letsencrypt_setup_extra_params: "--http-01-address {{ ansible_host }} --http-01-port 8888"
>>>> haproxy_ssl_letsencrypt_email: email at example.com <mailto:email at example.com>
>>>> haproxy_interval: 2000
>>>> 
>>>> user avatar user avatar 
>>>> haproxy_extra_services:
>>>>   # an internal only service for acme-challenge whose backend is certbot on the haproxy host
>>>>   - service:
>>>>       haproxy_service_name: letsencrypt
>>>>       haproxy_backend_nodes:
>>>>         - name: localhost
>>>>           ip_addr: {{ ansible_host }}                        #certbot binds to the internal IP
>>>>       backend_rise: 1                                        #quick rise and fall time for multinode deployment to succeed
>>>>       backend_fall: 2
>>>>       haproxy_bind:
>>>>         - 127.0.0.1                                          #bind to 127.0.0.1 as the local internal address  will be used by certbot
>>>>       haproxy_port: 8888                                     #certbot is configured with http-01-port to be 8888
>>>>       haproxy_balance_type: http
>>>> ==================================================================================
>>>> 
>>>> Yet, Horizon config for HAproxy is already defined in the default vars (https://github.com/openstack/openstack-ansible/blob/stable/ussuri/inventory/group_vars/haproxy/haproxy.yml <https://github.com/openstack/openstack-ansible/blob/stable/ussuri/inventory/group_vars/haproxy/haproxy.yml>) and we don’t know where ta add the required ACL to redirect the traffic from 80 port to 8888:
>>>> 
>>>> ====================================
>>>> haproxy_frontend_acls:                                 #use a frontend ACL specify the backend to use for acme-challenge
>>>>   letsencrypt-acl:
>>>>     rule: "path_beg /.well-known/acme-challenge/"
>>>>     backend_name: letsencrypt
>>>> ====================================
>>>> 
>>>> We know that this is fixed in OpenStack Ansible Victoria. Is it possible with Ussuri tho ?
>>>> 
>>>> Many thanks,
>>>> Best,
>>>> Marc-Antoine Godde
>>>> 
>>>> 
>> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20220222/1e5a1905/attachment-0001.htm>