Re: [openstack-dev] [kolla] Continued support of Fedora as a base platform

2016-06-30 Thread Haïkel 
2016-06-30 14:07 GMT+02:00 Steven Dake (stdake) :
> What really cratered our implementation of fedora was the introduction of
> DNF.  Prior to that, we led with Fedora.  I switched my focus to something
> slower moving (CentOS) so I could focus on a properly working RDO rather
> then working around the latest and greatest changes.
>
> That said, if someone wants to fix Kolla to run against dnf, that would be
> fantastic, as it will need to be done for CentOS8 an RHEL8.
>
> Regards
> -steve
>

That's something that we fixed for Fedora Cloud image. I'll give it a shot.

Regards,
H.

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [kolla] Continued support of Fedora as a base platform

2016-06-30 Thread Steven Dake (stdake) 
What really cratered our implementation of fedora was the introduction of
DNF.  Prior to that, we led with Fedora.  I switched my focus to something
slower moving (CentOS) so I could focus on a properly working RDO rather
then working around the latest and greatest changes.

That said, if someone wants to fix Kolla to run against dnf, that would be
fantastic, as it will need to be done for CentOS8 an RHEL8.

Regards
-steve

On 6/29/16, 6:07 PM, "Gerard Braad"  wrote:

>Hi,
>
>
>Kolla has supported Fedora AFAIK since the project started, and offers
>several other valid options:
>
>  # Valid options are [ centos, fedora, oraclelinux, ubuntu ]
>  #kolla_base_distro: "centos"
>
>but in recent time, it came to my attention that the support of Fedora
>is lacking. There could be several reasons for this;
>
>  1. interest
>  2. lack of resources
>  3. life cycle
>
>Firstly, might this be related to the fact that deploying on Fedora is
>not of interest to most? The majority of deployments of OpenStack
>happen on either Ubuntu or RHEL/CentOS. However, supporting Fedora
>early can help the deliverance on future version of RHEL/CentOS
>(although, there can be years in between before this happens). It is
>therefore still of importance.
>
>Second, and this is probably more likely the case. The Kolla project
>lacks the resources to maintain releasing on Fedora. Especially, since
>Fedora carries newer versions of software, there is a tendency of
>breakage. Automated tests is because of this of high importance.
>
>Third, since Fedora does not have a concept of Long-term releases, the
>release is only supported for a period of approximately 13 months.
>This is detailed in the Release Life Cycle [1] and EOL status page
>[2]. This means that after a release, like currently F24, the previous
>version like F22 will be phased out.
>
>A recent bugreport [3] about the image availability got resolved by
>implementing F22 (which would have been phased out just a month or two
>from now). The suggestion was to use CentOS for this. Maybe in this
>case it was... but should we?
>
>The question is not "Do we want to support Fedora?", but "Can we
>support Fedora?. If my time allows, I will certainly work on making
>this happen. But before, it might be needed to collect some of the
>feedback what has been done, what needs to be done... and what is
>currently the impediment of making it happen, like issues with
>versions of the dependencies.
>
>Would like to hear your thoughts...
>
>regards,
>
>
>Gerard
>
>[1]  https://fedoraproject.org/wiki/Fedora_Release_Life_Cycle
>[2]  https://fedoraproject.org/wiki/End_of_life
>[3]  https://bugs.launchpad.net/kolla/+bug/1589770
>
>-- 
>
>   Gerard Braad | http://gbraad.nl
>   [ Doing Open Source Matters ]
>
>__
>OpenStack Development Mailing List (not for usage questions)
>Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
>http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [kolla] Continued support of Fedora as a base platform

2016-06-30 Thread Haïkel 
My opinion as one of RDO release wranglers is not to support Fedora
for anything else that isn't trunk.
It's proven really hard to maintain all dependencies in a good state,
and when we managed to do that,
an update could break things at any time (like python-pymongo update
who was removed because of Pulp developers).

RDO actually ensure that spec files are buildable on Fedora but you'd
have to maintain dependencies separately and rely on
tools like yum priorities plugin to override base packages.

Fedora lifecycle is also not sync-ed with OpenStack, OpenStack being
released around 2 months before the next Fedora Stable.
So in practice, if you use stable N-1, you have 9 months of support
from Fedora, and updating to stable N requires some amount of work.

Regards,
H.

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

[openstack-dev] [kolla] Continued support of Fedora as a base platform

2016-06-29 Thread Gerard Braad 
Hi,


Kolla has supported Fedora AFAIK since the project started, and offers
several other valid options:

  # Valid options are [ centos, fedora, oraclelinux, ubuntu ]
  #kolla_base_distro: "centos"

but in recent time, it came to my attention that the support of Fedora
is lacking. There could be several reasons for this;

  1. interest
  2. lack of resources
  3. life cycle

Firstly, might this be related to the fact that deploying on Fedora is
not of interest to most? The majority of deployments of OpenStack
happen on either Ubuntu or RHEL/CentOS. However, supporting Fedora
early can help the deliverance on future version of RHEL/CentOS
(although, there can be years in between before this happens). It is
therefore still of importance.

Second, and this is probably more likely the case. The Kolla project
lacks the resources to maintain releasing on Fedora. Especially, since
Fedora carries newer versions of software, there is a tendency of
breakage. Automated tests is because of this of high importance.

Third, since Fedora does not have a concept of Long-term releases, the
release is only supported for a period of approximately 13 months.
This is detailed in the Release Life Cycle [1] and EOL status page
[2]. This means that after a release, like currently F24, the previous
version like F22 will be phased out.

A recent bugreport [3] about the image availability got resolved by
implementing F22 (which would have been phased out just a month or two
from now). The suggestion was to use CentOS for this. Maybe in this
case it was... but should we?

The question is not "Do we want to support Fedora?", but "Can we
support Fedora?. If my time allows, I will certainly work on making
this happen. But before, it might be needed to collect some of the
feedback what has been done, what needs to be done... and what is
currently the impediment of making it happen, like issues with
versions of the dependencies.

Would like to hear your thoughts...

regards,


Gerard

[1]  https://fedoraproject.org/wiki/Fedora_Release_Life_Cycle
[2]  https://fedoraproject.org/wiki/End_of_life
[3]  https://bugs.launchpad.net/kolla/+bug/1589770

-- 

   Gerard Braad | http://gbraad.nl
   [ Doing Open Source Matters ]

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev