Open Stack

On Tue, 26 Nov 2019 at 11:30, Jonathan Rosser
<jonathan.rosser at rd.bbc.co.uk> wrote:
>
>
>
> On 25/11/2019 17:48, Mark Goddard wrote:
> > Hi,
> >
> > During the recent kolla PTG discussions [1], we covered CentOS 8 in
> > some detail. For kolla there are several aspects to CentOS 8 support:
> >
> > * kolla-build execution host
> > * container images
> > * kolla-ansible execution host
> > * kolla-ansible remote hosts
> >
>
> Hi Mark,
>
> Whilst the container approach in openstack-ansible is different (LXC)
> and presents it's own challenges for Centos8, there is no doubt a common
> set of issues. We also do a container-less bare-metal deploy mode which
> would be the first to try to get working.
>
> I have a WIP patch for OSA for Centos-8 for a while now which really
> does not get so far.
>
> https://review.opendev.org/#/c/689629
>
> I've not yet managed to create dummy interfaces with network-manager
> which leads to a horrible hack setting up the networking for CI.
>
> Also failing was building of dbus-python which is required for the
> Ansible nmcli module, due to an incompatibility with autotools 1.16.
>
> There are a also few pieces we take from EPEL (lsyncd) which are missing.
>
> Then there is/was the absence of an obvious source of RDO packages.
>
> I am concerned about the difficulty of creating a tractable upgrade path
> which separates OS from OpenStack upgrades. For openstack-ansible, Rocky
> was the transition release which supported both Xenial and Bionic.
> Achieving the same transition release for Centos 7 to 8 upgrades looks
> challenging.
>
> Jon.

Thanks Jon, there's some useful information in there. Given that RDO
Ussuri will not support CentOS 7, I think you will also need to do the
7 to 8 transition on Train. The timing's not ideal here - I can see
all the deployment tools either needing to backport large changes to
stable/train, or delay their releases (if they haven't released yet).

Let's keep this thread going as we each hit different roadblocks.