[openstack-dev] [kolla] [bifrost] bifrost container.

Steven Dake (stdake) stdake at cisco.com
Mon May 9 21:51:34 UTC 2016

From: "Mooney, Sean K" <sean.k.mooney at intel.com<mailto:sean.k.mooney at intel.com>>
Reply-To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Date: Monday, May 9, 2016 at 11:03 AM
To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Subject: Re: [openstack-dev] [kolla] [bifrost] bifrost container.


If we choose to use bifrost to deploy ironic standalone I think combining kevins previous
suggestion of modifying the bifrost install playbook with Steve Dake’s suggestion of creating a series
of supervisord configs for running each of the service is a reasonable approch.

I am currently look to scope how much effort would be required to split  the main task in the bifrost-ironic-install role
into 3 files which would be included in the main.yml:
Install_componets.yml (executed when skip_install is not defiend)
Bootstrap_components.yml (executed when skip_bootstrap is not defiend)
Start_components.yml  (executed when skip_start is not defiend)
By default all three would be executed maintain the current behavior of bifrost today,.


During the kolla build of the biforst image the https://github.com/openstack/bifrost/blob/master/playbooks/install.yaml would be in
run with skip_bootstrap and skip_start defined as true so only Install_componets.yml will be executed by the main task.
This would install all software components of bifrost/ironic without preforming configuration or starting the services.


At deployment time during the bootstrap phase we would spawn an instance of the biforst-base container and invoke
https://github.com/openstack/bifrost/blob/master/playbooks/install.yaml with skip_install and skip_start defined executing Bootstrap_components.yml


Bootstrap_components.yml would encapsulate all logic related to creating the ironic db(running migration scripts) and generating the configuration
Files for the biforst components.

Finally in the start phase we have 3 options

a)      Spawn an instance of the bifrost-supervisor container and use supervisord to run the bifrost/ironic services (fat container)

b)      Spawn an instance of the bifrost-base container and Invoke https://github.com/openstack/bifrost/blob/master/playbooks/install.yaml with
skip_install and skip_bootstrap and allow biforst to star the services.(fat container)

c)       Spawn a series of containers each running a single service sharing the required volumes to allow them  to communicate (app containers)

A is path of least resistance (in a container).  I'd highly recommend doing this work in the upstream Bifrost community.  C is obviously where we want to be long term, but lets see where we can get to from A.  I'd be happy with A in general even if that was the last step we made in the progress of Bifrost & containerization of ironic.


I would welcome any input for the bifrost community on this especially related to the decomposition of the main.yml into 3 phases.
Im hoping to do a quick poc this week to see how easy it is to do this decomposition.

I would also like to call out upfront that depending on the scope of this item I may have to withdraw from contributing to it.
I work in intel’s network platforms group so enabling baremetal installation is somewhat outside the standard
Work our division undertakes.  If we can reuse bifrost to do most of the heavy lifting of creating the bifrost container and deploying ironic then
The scope of creating the bifrost container is small enough that I can justify spending some of my time working on it. if it requires
Significant changes to bifrost or rework of kolla’s ironic support then I will have to step back and focus more on feature that are closer aligned to
Our teams core networking and orchestration focus such as enhancing kolla to be able to deploy ovs with dpdk  and/or opendaylight which are
Also items I would like to contribute to this cycle. I don’t want to commit to delivering this feature unless I know I will have the time to work on
It but am happy to help where I can.

@kevin some replies to your questions inline.


From: Fox, Kevin M [mailto:Kevin.Fox at pnnl.gov]
Sent: Friday, May 6, 2016 9:17 PM
To: OpenStack Development Mailing List (not for usage questions) <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Subject: Re: [openstack-dev] [kolla] [bifrost] bifrost container.

I was under the impression bifrost was 2 things, one, an installer/configurator of ironic in a stand alone mode, and two, a management tool for getting machines deployed without needing nova using ironic.
[Mooney, Sean K] yes this is correct, bifrost does provide both install playbooks for deploying ironic in standalone mode and a series of playbooks for dynamically enrolling node in ironic and dynamically deploy imanges to host
Without requiring nova. Bifrost also provides intergration with Disk image builder to generate machine images if desired.

The first use case seems like it should just be handled by enhancing kolla's ironic container stuff to directly to handle the use case, doing things the kolla way. This seems much cleaner to me. Doing it at runtime looses most of the benefits of doing it in a container at all.
[Mooney, Sean K] I was not suggestiong doing the installation at runtime. Option 2 and 3   suggested spawning a container as part of the build in which the install playbook would be run.
That container would then be stopped and exported to form the base image for the bifrost continer(s). The base image (bifrost-postinstall)  would either be use to create  a fat containter using an init system such as supervisord to run each of the services
or be used as the base image for a set of bifrost container each of which ran a single component.

The second adds a lot of value I think, and thats what the bifrost container should be?
[Mooney, Sean K]  yes it does and I think it can be reused regarless or how we decide to deploy ironic.

From: Mooney, Sean K [sean.k.mooney at intel.com<mailto:sean.k.mooney at intel.com>]
Sent: Friday, May 06, 2016 12:54 PM
To: OpenStack Development Mailing List (not for usage questions)
Subject: Re: [openstack-dev] [kolla] [bifrost] bifrost container.

From: Steven Dake (stdake) [mailto:stdake at cisco.com]
Sent: Friday, May 6, 2016 6:56 PM
To: OpenStack Development Mailing List (not for usage questions) <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Subject: Re: [openstack-dev] [kolla] [bifrost] bifrost container.


Thanks for taking this on :)  I didn't know you had such an AR :)
[Mooney, Sean K] well if other want to do the work that ok with me too but I was planning on deploying bifrost
At home again anyway so I taught I  might as well try to automate the process while im at it.

From: "Mooney, Sean K" <sean.k.mooney at intel.com<mailto:sean.k.mooney at intel.com>>
Reply-To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Date: Friday, May 6, 2016 at 10:14 AM
To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Subject: [openstack-dev] [kolla] [bifrost] bifrost container.

Hi everyone.

Following up on my AR from the kolla host repository session
I started working on creating a kolla bifrost container.

Are some initial success it have hit a roadblock with the current install playbook provided by bifrost.
In particular the install playbook both installs the ironic dependencies and configure and runs the services.

What I'd do here is ignore the install playbook and duplicate what it installs.  We don't want to install at run time, we want to install at build time.  You weren't clear if that is what your doing.
[Mooney, Sean K] that is certainly an option but bifrost is an installer for ironic and its supporting service. Not using its installation scripts significantly reduces the value of
Integrating with bifrost vs fixing the existing ironic support in kolla and using that to provision the undercloud.

The reason we would ignore the install playbook is because it runs the services.  We need to run the services in a different way.  This will (as we discussed at ODS) be a fat container on the underlord cloud – which I guess is ok.  I'd recommend not using systemd, as that will break systemd systems badly.  Instead use a different init system, such as supervisord.
[Mooney, Sean K] if we don’t use the  bifrost install playbook then yes supervisord would be a good choice for the init system.
Looking at the official centos docker image https://hub.docker.com/_/centos/  they do provided instruction for running systemd containers tough I have had issues with this in the past.

The installation of ironic and its dependencies would not be a problem but the ansible service module is not cable able of starting the
Infrastructure services (mysql,rabbit …) without a running init system which is not present during the docker build.

When I created a biforst container in the past is spawned a Ubuntu upstart container then docker exec into the container and ran
Bifrost install script. This works because the init system is running and the service module could test and start the relevant services.

This leave me with 3 paths forward.

1.       I can continue to try and make the bifrost install script work with the kolla build system by using sed to modify the install playbook or try start systemd during the docker build.

2.       I can use the kolla build system to build only part of the image

a.        the bifrost-base image would be build with the kolla build system without running the bifrost playbook. This
would allow the existing allow the existing features of the build system such as adding headers/footers to be used.

b.      After the base image is built by kolla I can spawn an instance of bifrost-base with systemd running

c.       I can then connect to this running container and run the bifrost install script unmodified.

d.      Once it is finished I can stop the container and export it to an image “bifros-postinstall”.

e.      This can either be used directly (fat container) or as the base image for other container that run each of the ironic services (thin containers)

3.       I can  skip the kolla build system entirely and create a script/playbook that will build the bifrost container similar to 2.

Make a supervisord set of init scripts and make the docker file do what it was intended – install the files.  This is kind of a mashup of your 1-3 ideas.  Good thinking :)

While option 1 would fully use the kolla build system It is my least favorite as it is both hacky and complicated to make work.
Docker really was not designed to run systemd as part of docker build.

For option 2 and 3 I can provide a single playbook/script that will fully automate the build but the real question I have
Is should I use the kolla build system to make the base image or not.

If anyone else has suggestion on how I can progress  please let me know but currently I am leaning towards option 2.

If you have questions about my suggestion to use supervisord, hit me up on IRC.  Ideally we would also contribute these init scripts back into bifrost code base assuming they want them, which I think they would.  Nobody will run systemd in a container, and we all have an interest in seeing BiFrost as the standard bare metal deployment model inside or outside of containers.
[Mooney, Sean K]  I have briefly used supervisord before for a pet project https://github.com/SeanMooney/docker-devstack to create a container for running devstack so it did hot pollute my host.
supervisord  is a nice tool. Im just about to head home for the weekend but I might grab you on irc on Monday to follow up.


The only other option I see would be to not use a container and either install biforst on the host or in a vm.

GROAN – one advantage containers provide us is not mucking up the host OS with a bajillion dependencies.  I'd like to keep that part of Kolla intact :)
[Mooney, Sean K]  yes I would prefer not to break that too. This was basically the option of we don’t actually do the integration and instead just tell
The user how to use bifrost to do the deployment but leave it up to them do decide how to install it. so for me that was plan Z so we have a couple of letter
Go through first.

These would essentially be a no op for kolla as we would simply have to document how to install bifrost which is covered
Quite well as part of the bifrost project.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160509/93ccd391/attachment.html>

More information about the OpenStack-dev mailing list