10:28 a.m.
Greetings,
A bug [1] was raised in launchpad and I wanted to get the communities input
on it. The background is that atm we only CI TripleO with CentOS and RHEL
and have documented these two distributions are the ONLY two that have
community support [2].
Using RDO-Cloud we now have the ability to CI with Fedora. So the question
is that something that has enough demand and do we have the resources to CI
it and provide community support?
Let me know what you think.
[1] https://bugs.launchpad.net/tripleo/+bug/1740928
[2]https://docs.openstack.org/tripleo-quickstart/latest/readme.html
11:27 a.m.
On Fri, Jan 12, 2018 at 4:28 PM, Wesley Hayutin <whayutin(a)redhat.com> wrote:
Greetings,
A bug [1] was raised in launchpad and I wanted to get the communities input
on it. The background is that atm we only CI TripleO with CentOS and RHEL
and have documented these two distributions are the ONLY two that have
community support [2].
Using RDO-Cloud we now have the ability to CI with Fedora. So the question
is that something that has enough demand and do we have the resources to CI
it and provide community support?
Let me know what you think.
I think it's important to be specific here - AFAICS the bug is asking
for *tripleo-quickstart* support of Fedora "at least as the primary
hypervisor", not necessarily for all of TripleO to support Fedora?
I think quickstart support of Fedora as the test node/hypervisor is
desirable (FWIW I used to test like this regularly back when
instack-virt-setup was supported and it worked fine).
Personally I don't think we have the resources (or demand) to support
all of TripleO on Fedora at this point - I think we already have
enough pending work improving our stability and test coverage just on
CentOS in upstream CI.
Thanks,
Steve
12:30 p.m.
On Fri, Jan 12, 2018 at 12:27 PM, Steven Hardy <shardy(a)redhat.com> wrote:
On Fri, Jan 12, 2018 at 4:28 PM, Wesley Hayutin
<whayutin(a)redhat.com>
wrote:
> Greetings,
>
> A bug [1] was raised in launchpad and I wanted to get the communities
input
> on it. The background is that atm we only CI TripleO with CentOS and
RHEL
> and have documented these two distributions are the ONLY two that have
> community support [2].
>
> Using RDO-Cloud we now have the ability to CI with Fedora. So the
question
> is that something that has enough demand and do we have the resources to
CI
> it and provide community support?
>
> Let me know what you think.
I think it's important to be specific here - AFAICS the bug is asking
for *tripleo-quickstart* support of Fedora "at least as the primary
hypervisor", not necessarily for all of TripleO to support Fedora?
I suspect one could lead to the other in time.
I think quickstart support of Fedora as the test node/hypervisor is
desirable (FWIW I used to test like this regularly back when
instack-virt-setup was supported and it worked fine).
Things may work from time to time, however our policy is that if it's not
in CI it's not supported.
Personally I don't think we have the resources (or demand) to support
all of TripleO on Fedora at this point - I think we already have
enough pending work improving our stability and test coverage just on
CentOS in upstream CI.
Thanks,
OK.. good to know
Thanks Steve
Steve
12:32 p.m.
Hi,
from chatting to a guy who does work with fedora and supports bugs etc. His
advice is that because of the fast cycle of fedora releases, they are not
suitable for production machines. As such, I'd concur with sticking with
RHEL / CentOS.
Regards,
Phill.
On 12 January 2018 at 17:27, Steven Hardy <shardy(a)redhat.com> wrote:
On Fri, Jan 12, 2018 at 4:28 PM, Wesley Hayutin
<whayutin(a)redhat.com>
wrote:
> Greetings,
>
> A bug [1] was raised in launchpad and I wanted to get the communities
input
> on it. The background is that atm we only CI TripleO with CentOS and
RHEL
> and have documented these two distributions are the ONLY two that have
> community support [2].
>
> Using RDO-Cloud we now have the ability to CI with Fedora. So the
question
> is that something that has enough demand and do we have the resources to
CI
> it and provide community support?
>
> Let me know what you think.
I think it's important to be specific here - AFAICS the bug is asking
for *tripleo-quickstart* support of Fedora "at least as the primary
hypervisor", not necessarily for all of TripleO to support Fedora?
I think quickstart support of Fedora as the test node/hypervisor is
desirable (FWIW I used to test like this regularly back when
instack-virt-setup was supported and it worked fine).
Personally I don't think we have the resources (or demand) to support
all of TripleO on Fedora at this point - I think we already have
enough pending work improving our stability and test coverage just on
CentOS in upstream CI.
Thanks,
Steve
_______________________________________________
dev mailing list
dev(a)lists.rdoproject.org
http://lists.rdoproject.org/mailman/listinfo/dev
To unsubscribe: dev-unsubscribe(a)lists.rdoproject.org
4:27 p.m.
On Fri, Jan 12, 2018 at 1:32 PM, Phill. Whiteside <phillwuk(a)gmail.com> wrote:
Hi,
from chatting to a guy who does work with fedora and supports bugs etc. His
advice is that because of the fast cycle of fedora releases, they are not
suitable for production machines. As such, I'd concur with sticking with
RHEL / CentOS.
By that logic, you'd avoid OpenStack too, since they follow almost
exactly the same schedule and cadence. In general, I've been rather disappointed
about the lack of effort in testing OpenStack on top of Fedora, as
Fedora is the *upstream* of RHEL/CentOS, and there's no good reason
for not doing it.
I would not be surprised if RDO OpenStack is not going to be ready for
Python 3 because there's no work going on to at least build against
Fedora. And that has nasty trickle-down effects...
--
真実はいつも一つ！/ Always, there's only one truth!
6:43 a.m.
On 12/01/2018 23:27, Neal Gompa wrote:
On Fri, Jan 12, 2018 at 1:32 PM, Phill. Whiteside
<phillwuk(a)gmail.com> wrote:
> Hi,
>
> from chatting to a guy who does work with fedora and supports bugs etc. His
> advice is that because of the fast cycle of fedora releases, they are not
> suitable for production machines. As such, I'd concur with sticking with
> RHEL / CentOS.
>
By that logic, you'd avoid OpenStack too, since they follow almost
exactly the same schedule and cadence. In general, I've been rather disappointed
about the lack of effort in testing OpenStack on top of Fedora, as
Fedora is the *upstream* of RHEL/CentOS, and there's no good reason
for not doing it.
Being short handed is a good reason, there's a SIG to help fill that gap
partially though.
I would not be surprised if RDO OpenStack is not going to be ready
for
Python 3 because there's no work going on to at least build against
Fedora. And that has nasty trickle-down effects...
Well, actually we do have a trunk builder on Fedora and we regularly fix
build issues. Most libs and clients are already python3 compliant.
Regards,
H.
8:09 a.m.
On Sat, Jan 13, 2018 at 7:43 AM, Haïkel Guémar <hguemar(a)redhat.com> wrote:
On 12/01/2018 23:27, Neal Gompa wrote:
>
> On Fri, Jan 12, 2018 at 1:32 PM, Phill. Whiteside <phillwuk(a)gmail.com>
> wrote:
>>
>> Hi,
>>
>> from chatting to a guy who does work with fedora and supports bugs etc.
>> His
>> advice is that because of the fast cycle of fedora releases, they are not
>> suitable for production machines. As such, I'd concur with sticking with
>> RHEL / CentOS.
>>
>
> By that logic, you'd avoid OpenStack too, since they follow almost
> exactly the same schedule and cadence. In general, I've been rather
> disappointed
> about the lack of effort in testing OpenStack on top of Fedora, as
> Fedora is the *upstream* of RHEL/CentOS, and there's no good reason
> for not doing it.
>
Being short handed is a good reason, there's a SIG to help fill that gap
partially though.
I'm not sure I believe the "short-handed" bit. Care to elaborate on
this? The RDO project is fairly big with quite a few people who work
on this...
My understanding is that the OpenStack SIG in Fedora is not supposed
to work within the Fedora SCM workflow and is supposed to work through
the RDO one, so I would be shocked if anyone actually wanted to help
in that regard (not that many people want to put themselves through
Gerrit and the unusual workflow of RDO when *nothing else* in Fedora
works that way). Also, I'm not sure it's even allowed to force that in
a SIG, since the rules explicitly state that Fedora Dist-Git is the
canonical location for packaging sources and where things should be
done.
Even I personally don't want to touch any openstack client/lib
packages in Fedora since I don't want to risk the ire of people who
use it as a mirror from RDO Git just to build Fedora packages.
> I would not be surprised if RDO OpenStack is not going to be
ready for
> Python 3 because there's no work going on to at least build against
> Fedora. And that has nasty trickle-down effects...
>
Well, actually we do have a trunk builder on Fedora and we regularly fix
build issues. Most libs and clients are already python3 compliant.
But you're not proactively adapting things for new Fedora packaging
and stuff that doesn't exist in RHEL (or even EPEL, as now Python 3 in
EPEL is transitioning to Python 3.6), right?
Are you also helping to move OpenStack as a platform to Python 3? Last
I checked, it didn't seem like that was the case, because you guys don't do
work for anything RHEL 7 doesn't have. :(
--
真実はいつも一つ！/ Always, there's only one truth!
2506
days inactive
2507
days old
6 comments
5 participants
participants (5)
-
Haïkel Guémar -
Neal Gompa -
Phill. Whiteside -
Steven Hardy -
Wesley Hayutin