Contributors mailing list archives

contributors@odoo-community.org

Re: Proposal for new repo - Clouder

by
LasLabs, Dave Lasley
- 14/10/2016 13:44:50
It definitely seems like a lot of us have been working towards the same goal for quite some time now. 

We might be expanding our scope of this repo a bit too much if we start including Ansible/Chef/Puppet as well. IMO these are related to the underlying server spin-ups & management, and should be contained within a repo of their own. `Infrastructure-inventory` I believe fits this purpose in terms of naming of the inventory files. Alternatively, `infrastructure-configuration` or `infrastructure-puppeteer` may better describe the actions performed by the modules within the repo.

I think we’re digressing a bit though, as the intent here is specifically to manage container infrastructure and the applications underneath. In a best-practice scenario, containers should be completely naive of the server infrastructure underneath them. We should enforce this practice by isolating the configuration of servers from the configuration of containers.

FYI we have started a roadmap in the Clouder repo & will be beginning an incubator branch for some rather drastic changes. 

It seems to me that this email thread has enough activities and :+1s to support an OCA repo, so I propose that the completion of the incubator is a good time for migration into a new OCA repo named `infrastructure-clouder` if we want to align with our infrastructure prefixes or simply `clouder` to align with history of the modules.

On Oct 13, 2016, at 1:53 AM, Jos De Graeve <jos.degraeve@apertoso.be> wrote:

Hi,

At Apertoso we also built something focused on the way we do development and hosting of our customer instances.


We published what we use internally here:


What i see today is that everybody already uses something to manage their infrastructure, but everybody is doing it in their own way   There are a lot of ways to do your infrastructure wrong, but there are also multiple ways to do it right.  Bottom line is that a lot of duplicated work already has been done.

What everybody needs is the basic host inventory information on which instances you need to run, and their configuration parameters.  This inventory information can be on odoo instances, but imho it should not be limited only to that.

The next step is to use that information to setup your infrastructure.  If you use a classical setup where instances are deployed on a VM, one can use ansible for that.  Ansible allows to use a "dynamic inventory" script.  An example for such a script is: https://github.com/apertoso/tools/blob/master/ansible_env_all.py. Next to ansible there are other tools as well. ( Puppet, Chef, ... )

For the hipster dev-ops there is docker.  With docker your instances get simplified to the absolute minimal, so there is no longer a real need for a full fledged configuration management tool to setup an individual instance.  We created a few scripts to take our inventory information and build a docker image for that.  The scripts fetch the required modules, generate a dockerfile and run docker to build an image. These images can be used on a production environment as well as on your development laptop as well as on your staging environment.

What @YannickB has done is great work.  Unfortunately i could not attend his presentation last week but i'm looking forward to see the slides of his presentation.  His clouder project is the best i seen so far.  What i don't like however, is that i see templating code inside an odoo module.  So there is no de-coupling between the "infrastructure data" and the "infrastructure implementation". 

So i feel there is real opportunity here to put a lot of experience together and build something great :)

As a suggestion i would like to propose "infrastructure-inventory" 

regards,

Jos De Graeve - Apertoso business ICT


Guido Gezellelaan 16 - B-9800 Deinze - Belgium
 
Direct: +32 9 381 64 51
General:+32 9 381 64 50
Mobile: +32 475 54 68 80
mail/im/skype: Jos.DeGraeve@apertoso.be - apertoso

2016-10-12 14:22 GMT+02:00 Juan José Scarafía (ADHOC) <scarafia.juanjose@gmail.com>:
+1 for "clouder". FYI we have a set of modules (https://github.com/ingadhoc/odoo-infrastructure) for a similar approach but only focus for odoo deployment. Ther are not OCA quality modules. They need a complete refactoring. I think I like Yennik clouder approach more, let's join efforts. 

El sáb., 8 oct. 2016 a las 15:07, Daniel Reis (<dgreis@sapo.pt>) escribió:
I also agree with "clouder":
This is a project with it's own identity and it wouldn't be helpful to 
later add here unrelated modules just because they are features for SaaS 
deployment.

--
Daniel


_______________________________________________
Mailing-List: http://odoo-community.org/groups/contributors-15
Post to: mailto:contributors@odoo-community.org
Unsubscribe: http://odoo-community.org/groups?unsubscribe

--

Ing. Juan José Scarafía

(+54 9 341)153 278039
skype: jjscarafia
twitter: @jjscarafia
github: @jjscarafia
image-2015-09-10.png

_______________________________________________
Mailing-List: http://odoo-community.org/groups/contributors-15
Post to: mailto:contributors@odoo-community.org
Unsubscribe: http://odoo-community.org/groups?unsubscribe