[Openstack-operators] Ops Meetups Team - meeting minutes from today's IRC meeting

2018-06-05 Thread Chris Morgan
We had a brief meeting today, minutes linked below. Topic included
restarting the operator-related docs, prep for Denver PTG and some
organisational things for the team.

We'll meet bi-weekly at 10am EST for the time being, so next meeting is
2018-6-19 at 10 am EST

Minutes:
http://eavesdrop.openstack.org/meetings/ops_meetup_team/2018/ops_meetup_team.2018-06-05-14.25.html
10:52 AM Minutes (text):
http://eavesdrop.openstack.org/meetings/ops_meetup_team/2018/ops_meetup_team.2018-06-05-14.25.txt
10:52 AM Log:
http://eavesdrop.openstack.org/meetings/ops_meetup_team/2018/ops_meetup_team.2018-06-05-14.25.log.html

-- 
Chris Morgan 
___
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators


[Openstack-operators] [FEMDC] meetings suspended until further notice

2018-06-05 Thread lebre . adrien
Dear all,

Following the exchanges we had during the Vancouver summit, in particular the 
non-sense to maintain/animate two groups targeting similar challenges (ie., the 
FEMDC SIG [1] and the new Edge Computing Working group [2]), FEMDC meetings are 
suspended until further notice.  

If you are interested by Edge Computing discussions, please see information 
available on the new edge wiki page [2]. 

Thanks
ad_ri3n_

[1]https://wiki.openstack.org/wiki/Fog_Edge_Massively_Distributed_Clouds
[2]https://wiki.openstack.org/wiki/Edge_Computing_Group

___
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators


Re: [Openstack-operators] [openstack-ansible][releases][governance] Change in OSA roles tagging

2018-06-05 Thread Doug Hellmann
Excerpts from Jean-Philippe Evrard's message of 2018-06-05 10:14:12 +0200:
> Hello,
> 
> *TL:DR;* If you are an openstack-ansible user, consuming our roles directly,
> with tags, without using openstack-ansible plays or integrated repo,
> then things will change for you. Start using git shas instead of tags.
> All other openstack-ansible users should not see a difference, even if
> they use openstack-ansible tags.
> 
> 
> During the summit, I had a discussion with dhellman (and smcginnis)
> to change how openstack-ansible does its releases.
> 
> Currently, we tag openstack-ansible + many roles under our umbrella
> every two weeks. As far as I know, nobody requested to have roles
> tagged every two weeks. Only OpenStack-Ansible need to be tagged
> for consumption. Even people using our roles directly outside
> openstack-ansible generally use sha for roles. We don't rely on
> ansible galaxy.
> 
> Because there is no need to tag the roles, there is no need to make them
> part of the "openstack-ansible" deliverable [1][2]. I will therefore
> clarify the governance repo for that, separating the roles, each of them
> with their own deliverable, instead of grouping some roles within
> openstack-ansible, and some others outside it.
> 
> With this done, a release of openstack-ansible becomes straightforward
> using the standard release tooling. The releases of openstack-ansible
> becomes far simpler to request, review, and will not have timeouts
> anymore :p
> 
> There are a few issues I see from the change. Still according to the
> discussion, it seems we can overcome those.
> 
> 1. As this will be applied on all the branches, we may reach some
> issues with releasing in the next days. While the validate tooling
> of releases has shown me that it wouldn't be a problem (just
> warning) to not have all the repos in the deliverable, I would
> expect a governance change could be impactful.
> However, that is only impacting openstack-ansible, releases,
> and governance team: Keep in mind, openstack-ansible will not
> change for its users, and will still be tagged as you know it.
> 
> 2. We will keep branching our roles the same way we do now. What
> we liked for roles being part of this deliverable, is the ability
> of having them automatically branched and their files adapted.
> To what I heard, it is still possible to do so, by having a
> devstack-like behavior, which branches on a sha, instead of
> branching on tag. So I guess it means all our roles will now be
> part of release files like this one [3], or even on a single release
> file, similar to it.

Right, you can set the stable-branch-type field to 'tagless' (see
http://git.openstack.org/cgit/openstack/releases/tree/README.rst#n462) and
then set the branch location field to the SHA you want to use.

If you would be ready to branch all of the roles at one time, you could
put all of them into 1 deliverable file. Otherwise, you will want to
split them up into their own files.

And since you have so many, I will point out that we're really into
automation over here on the release team, and if you wanted to work on
making the edit-deliverable command smart enough to determine the SHA
for you I could walk you through that code to get you started.

Doug

> 
> What I would like to have, from this email, is:
> 1. Raise awareness to all the involved parties;
> 2. Confirmation we can go ahead, from a governance standpoint;
> 3. Confirmation we can still benefit from this automatic branch
> tooling.
> 
> Thank you in advance.
> 
> Jean-Philippe Evrard (evrardjp)
> 
> 
> [1]: 
> https://github.com/openstack/governance/blob/8215c5fd9b464b332b310bbb767812fefc5d9174/reference/projects.yaml#L2493-L2540
> [2]: 
> https://github.com/openstack/releases/blob/9db5991707458bbf26a4dd9f55c2a01fee96a45d/deliverables/queens/openstack-ansible.yaml#L768-L851
> [3]: 
> https://github.com/openstack/releases/blob/9db5991707458bbf26a4dd9f55c2a01fee96a45d/deliverables/queens/devstack.yaml
> 

___
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators


Re: [Openstack-operators] Switching from Fuel to OpenStack-Ansible

2018-06-05 Thread Matthew John
Hi both,


Thanks for the info.


Yep, I've had a look at the reference architecture and will probably end up 
running three controllers with six compute nodes using Ceph for storage. 
Initial thoughts are to put the Ceph OSDs on the compute nodes which each have 
4x400GB SSD drives spare and use the controllers as the ceph monitors.


It is a month or two until I start the migration so will hop on to IRC if I 
have any issues or questions!


Cheers,

Matt

---
Dr Matt John
Engineer (Service Delivery - COMSC)
School of Computer Science & Informatics
Cardiff University, 5 The Parade, Cardiff, CF24 3AA
Tel: +44 2920 876536
joh...@cardiff.ac.uk
The University welcomes correspondence in Welsh or English. Corresponding in 
Welsh will not lead to any delay.

Dr Matt John
Peiriannydd (Cyflwyno Gwasanaeth - COMSC)
Ysgol Cyfrifiadureg a Gwybodeg
Prifysgol Caerdydd, 5 The Parade, Caerdydd, CF24 3AA
Ffôn : +44 2920 876536
joh...@caerdydd.ac.uk
Mae'r Brifysgol yn croesawu gohebiaeth yn Gymraeg neu'n Saesneg. Ni fydd gohebu 
yn Gymraeg yn creu unrhyw oedi.


From: Jean-Philippe Evrard 
Sent: 05 June 2018 09:14:52
To: Matthew John
Cc: openstack-operators@lists.openstack.org
Subject: Re: [Openstack-operators] Switching from Fuel to OpenStack-Ansible

Hello,

That's doable indeed! (In fact it's exactly what I did during Kilo
timeframe! :p)
Have you looked at the openstack-ansible deploy guide? It should help
you understand the process. On the way it should link you to the
reference architecture, where you can have more details about the
network flows.

Don't hesitate to join us on our IRC channel for more detailed
questions, on freenode #openstack-ansible.
If you want to continue by email, don't hesitate to put
[openstack-ansible] in the email title :)

Best regards,
Jean-Philippe Evrard (evrardjp)
___
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators


Re: [Openstack-operators] Switching from Fuel to OpenStack-Ansible

2018-06-05 Thread Jean-Philippe Evrard
Hello,

That's doable indeed! (In fact it's exactly what I did during Kilo
timeframe! :p)
Have you looked at the openstack-ansible deploy guide? It should help
you understand the process. On the way it should link you to the
reference architecture, where you can have more details about the
network flows.

Don't hesitate to join us on our IRC channel for more detailed
questions, on freenode #openstack-ansible.
If you want to continue by email, don't hesitate to put
[openstack-ansible] in the email title :)

Best regards,
Jean-Philippe Evrard (evrardjp)

___
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators


[Openstack-operators] [openstack-ansible][releases][governance] Change in OSA roles tagging

2018-06-05 Thread Jean-Philippe Evrard
Hello,

*TL:DR;* If you are an openstack-ansible user, consuming our roles directly,
with tags, without using openstack-ansible plays or integrated repo,
then things will change for you. Start using git shas instead of tags.
All other openstack-ansible users should not see a difference, even if
they use openstack-ansible tags.


During the summit, I had a discussion with dhellman (and smcginnis)
to change how openstack-ansible does its releases.

Currently, we tag openstack-ansible + many roles under our umbrella
every two weeks. As far as I know, nobody requested to have roles
tagged every two weeks. Only OpenStack-Ansible need to be tagged
for consumption. Even people using our roles directly outside
openstack-ansible generally use sha for roles. We don't rely on
ansible galaxy.

Because there is no need to tag the roles, there is no need to make them
part of the "openstack-ansible" deliverable [1][2]. I will therefore
clarify the governance repo for that, separating the roles, each of them
with their own deliverable, instead of grouping some roles within
openstack-ansible, and some others outside it.

With this done, a release of openstack-ansible becomes straightforward
using the standard release tooling. The releases of openstack-ansible
becomes far simpler to request, review, and will not have timeouts
anymore :p

There are a few issues I see from the change. Still according to the
discussion, it seems we can overcome those.

1. As this will be applied on all the branches, we may reach some
issues with releasing in the next days. While the validate tooling
of releases has shown me that it wouldn't be a problem (just
warning) to not have all the repos in the deliverable, I would
expect a governance change could be impactful.
However, that is only impacting openstack-ansible, releases,
and governance team: Keep in mind, openstack-ansible will not
change for its users, and will still be tagged as you know it.

2. We will keep branching our roles the same way we do now. What
we liked for roles being part of this deliverable, is the ability
of having them automatically branched and their files adapted.
To what I heard, it is still possible to do so, by having a
devstack-like behavior, which branches on a sha, instead of
branching on tag. So I guess it means all our roles will now be
part of release files like this one [3], or even on a single release
file, similar to it.

What I would like to have, from this email, is:
1. Raise awareness to all the involved parties;
2. Confirmation we can go ahead, from a governance standpoint;
3. Confirmation we can still benefit from this automatic branch
tooling.

Thank you in advance.

Jean-Philippe Evrard (evrardjp)


[1]: 
https://github.com/openstack/governance/blob/8215c5fd9b464b332b310bbb767812fefc5d9174/reference/projects.yaml#L2493-L2540
[2]: 
https://github.com/openstack/releases/blob/9db5991707458bbf26a4dd9f55c2a01fee96a45d/deliverables/queens/openstack-ansible.yaml#L768-L851
[3]: 
https://github.com/openstack/releases/blob/9db5991707458bbf26a4dd9f55c2a01fee96a45d/deliverables/queens/devstack.yaml

___
OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators