fuel-dev team mailing list archive

[Harsha Ramamurthy <hramamur@xxxxxxxxxxx>]

Hi Mike


1.    Thank you for the instructions to subscribe to the mailing list. I have now completed the subscription process. You should hopefully see this email posted on the list.

2.    Regarding the setup: here are the high level details for the two plugins that are going to be certified: VCS<https://wiki.openstack.org/wiki/Brocade-neutron-plugin> and ADX<https://wiki.openstack.org/w/images/8/86/OpenStack-Neutron-Brocade-LBaaS-BP.pdf> plugins. I think the model suggested below (1 Fuel + 1 Controller + 2 computes) seems good to validate the plugin functionality. This is the same setup we have used for certifying the plugin with another partner as well. I will, however, let you take a look at the plugin details and confirm.

3.    On the bug related to Brocade NICs – let me take a look, have some internal discussion and get back to you.

Regards
Harsha

From: Mike Scherbakov [mailto:mscherbakov@xxxxxxxxxxxx]
Sent: Thursday, April 24, 2014 4:36 AM
To: Harsha Ramamurthy
Cc: Anuj; rsrinivasan@xxxxxxxxxxxx; Prakash Kaligotla; fuel-dev@xxxxxxxxxxxxxxxxxxx; jpipes@xxxxxxxxxxxx; Didier Stolpe; Brian Imber (CW); Kenneth Ross; Easter David; adanin@xxxxxxxxxxxx
Subject: Re: Query on Mirantis Openstack deployment

Hi all,
first of all, you need to subscribe to the mailing list. Message didn't get posted there I believe: https://www.mail-archive.com/fuel-dev@xxxxxxxxxxxxxxxxxxx/
To subscribe, please register at launchpad.net<http://launchpad.net>, join https://launchpad.net/~fuel-dev team, and click on "Subscribe" button in Mailing List section (left bottom of the page).

Vladimir provided a good overview on how process should look like in general. To answer your question about setup, ideally I'd need to know more about integration: what kind of plugins, does it affect computes only (or controllers as well), etc.
We actually have a bug about Brocade NICs support: https://bugs.launchpad.net/fuel/+bug/1260492, and we certainly want to fix it - however we don't have hardware to test a fix on. If you can help, that would be great.

Regarding particular question on setup: I assume that 1 Fuel Master node + 1 controller + 2 computes are fine. 2 computes to test connectivity between them. However, again, I'd like to know more details what exactly is in the picture, what NICs or any specific features we want to make sure Fuel supports.

From certification standpoint for this particular deal (in terms of hardware & software) we have three layers of testing:

 1.  Fuel CI<https://fuel-jenkins.mirantis.com/> which runs a number of checks on every proposed change into Gerrit<https://review.openstack.org/#/q/project:%255Estackforge/fuel-.*,n,z>
 2.  System tests <http://docs.mirantis.com/fuel-dev/develop/env.html#running-the-fuel-system-tests>  which are currently ran on Mirantis hardware inside of Mirantis network
 3.  Manual testing, including acceptance testing, which is done by QA team
Ideally, we would love to follow OpenStack model here<http://ci.openstack.org/third_party.html>, and simply add your CI slave to fuel-jenkins master, so for every patch provided, we can get +1 or -1 depending on the result Brocade support by that particular patch.
Here we want to be no different than https://wiki.openstack.org/wiki/Neutron_Plugins_and_Drivers#Testing_Requirements for now, and instead of inventing a wheel, simply follow existing guidelines, and improve if necessary.
As far as I can see from the table of supported drivers in Neutron on that page, Brocade is already supported by Neutron:
Brocade

Shiv Haris

shivh

For example: https://review.openstack.org/#/c/84519/20, there is "Brocade CI" in list of reviewers. I'd think that it would be best for us to have a similar CI bot.
Then, we might want to have some system test(s), which we run every night. And finally, to cover Brocade support in manual acceptance testing.

We would be glad to help with setting up CI and anything required to make this work.
Please let me know what are your thoughts on this.

Thank you,

On Thu, Apr 24, 2014 at 12:36 AM, Harsha Ramamurthy <hramamur@xxxxxxxxxxx<mailto:hramamur@xxxxxxxxxxx>> wrote:
Raju

[cid:image008.jpg@01CF5F9C.7278F3E0]


We will follow the above model (taken from Mirantis documentation) but with a few changes. The setup we plan to have will include:


1.       2  Controller Nodes

2.       7 Compute Nodes

3.       1 Fuel Node

4.       We won’t have a Cinder Node as our plugin is focused on Networking

Does this sound good? If so, we will go ahead with the setup.

Thanks
Harsha



From: Harsha Ramamurthy
Sent: Wednesday, April 23, 2014 11:35 AM
To: 'Anuj'; 'rsrinivasan@xxxxxxxxxxxx<mailto:rsrinivasan@xxxxxxxxxxxx>'
Cc: Prakash Kaligotla; 'fuel-dev@xxxxxxxxxxxxxxxxxxx<mailto:fuel-dev@xxxxxxxxxxxxxxxxxxx>'; 'jpipes@xxxxxxxxxxxx<mailto:jpipes@xxxxxxxxxxxx>'; Didier Stolpe; Brian Imber (CW); Kenneth Ross; 'Easter David'; 'mscherbakov@xxxxxxxxxxxx<mailto:mscherbakov@xxxxxxxxxxxx>'; 'adanin@xxxxxxxxxxxx<mailto:adanin@xxxxxxxxxxxx>'

Subject: RE: Query on Mirantis Openstack deployment

Keeping Raju in the loop as discussed in the meeting today.

Rgds

From: Harsha Ramamurthy
Sent: Wednesday, April 23, 2014 10:42 AM
To: 'Anuj'
Cc: Prakash Kaligotla; fuel-dev@xxxxxxxxxxxxxxxxxxx<mailto:fuel-dev@xxxxxxxxxxxxxxxxxxx>; jpipes@xxxxxxxxxxxx<mailto:jpipes@xxxxxxxxxxxx>; Didier Stolpe; Brian Imber (CW); Kenneth Ross; Easter David; mscherbakov@xxxxxxxxxxxx<mailto:mscherbakov@xxxxxxxxxxxx>; adanin@xxxxxxxxxxxx<mailto:adanin@xxxxxxxxxxxx>

Subject: RE: Query on Mirantis Openstack deployment


1.    For partner certification is there a required/recommended deployment model that needs to be followed – such as number of Controller and compute nodes, separate node for Fuel etc. There is a suggested model for deployment in the documentation on Mirantis website – which seems to incorporate HA considerations, but is that necessary for our setup aimed at plugin certification?

2.    We are of the understanding that we will be receiving certain test scripts that need to be executed as part of certification process. How do we get hold of those?



Regards
Harsha

From: Anuj [mailto:akumar@xxxxxxxxxxxx]
Sent: Wednesday, April 23, 2014 10:33 AM
To: Harsha Ramamurthy
Cc: Prakash Kaligotla; fuel-dev@xxxxxxxxxxxxxxxxxxx<mailto:fuel-dev@xxxxxxxxxxxxxxxxxxx>; jpipes@xxxxxxxxxxxx<mailto:jpipes@xxxxxxxxxxxx>; Didier Stolpe; Brian Imber (CW); Kenneth Ross; Easter David; mscherbakov@xxxxxxxxxxxx<mailto:mscherbakov@xxxxxxxxxxxx>; adanin@xxxxxxxxxxxx<mailto:adanin@xxxxxxxxxxxx>
Subject: Re: Query on Mirantis Openstack deployment

It is the right forum. What are your questions?
Cheers!
AK.



On Apr 23, 2014, at 12:11 PM, Harsha Ramamurthy <hramamur@xxxxxxxxxxx<mailto:hramamur@xxxxxxxxxxx>> wrote:

Hi Folks,

Any updates? Is this the right forum to ask certification related questions? Any inputs would be greatly appreciated. Thanks..

Regards
Harsha


From: Prakash Kaligotla
Sent: Monday, April 21, 2014 2:36 PM
To: Harsha Ramamurthy; fuel-dev@xxxxxxxxxxxxxxxxxxx<mailto:fuel-dev@xxxxxxxxxxxxxxxxxxx>; jpipes@xxxxxxxxxxxx<mailto:jpipes@xxxxxxxxxxxx>
Cc: Didier Stolpe; Brian Imber (CW); Kenneth Ross; deaster@xxxxxxxxxxxx<mailto:deaster@xxxxxxxxxxxx>; mscherbakov@xxxxxxxxxxxx<mailto:mscherbakov@xxxxxxxxxxxx>; adanin@xxxxxxxxxxxx<mailto:adanin@xxxxxxxxxxxx>; akumar@xxxxxxxxxxxx<mailto:akumar@xxxxxxxxxxxx>
Subject: RE: Query on Mirantis Openstack deployment

Two other questions

1.       We are of the understanding that we will be receiving certain test scripts that need to be executed as part of certification process. How do we get hold of those?
2.       Is this the right forum to ask such questions about certification? If not, please redirect to appropriate folks.

Thanks
Prakash

From: Harsha Ramamurthy
Sent: Monday, April 21, 2014 2:25 PM
To: fuel-dev@xxxxxxxxxxxxxxxxxxx<mailto:fuel-dev@xxxxxxxxxxxxxxxxxxx>; jpipes@xxxxxxxxxxxx<mailto:jpipes@xxxxxxxxxxxx>
Cc: Prakash Kaligotla
Subject: Query on Mirantis Openstack deployment

Hi Folks,

We are currently working on setting up an environment for certification of Brocade networking plugins by Mirantis. Our current plan is to use 1 controller and 1 compute node but the model suggested in Mirantis documentation<http://docs.mirantis.com/fuel/fuel-4.1/pre-install-guide.html#preparing-the-mirantis-openstack-deployment> seems to be as below:

<image001.png>



For partner certification, is there a required/recommended deployment model that needs to be followed? Appreciate your inputs here. Thank you…

Regards
Harsha




--
Mike Scherbakov
#mihgen