Hi All,
Hope y'all had a great PTG last
week and a great weekend. I would like to thank everyone who
participated/contributed to Tripleo PTG sessions and helped all of us
"share and learn". Though personally I don't like long emails, this
summary is going to be a bit longer, please bear with me.
We
had a total of ten sessions spread over two days with very good
participation (most sessions with 35+ [max 44] in attendance) and we
shared/ discussed a multitude of topics.
# Welcome and Zed Retrospective
The
highlights of the Zed cycle was collaboration across teams to add some
new features and fine-tune features implemented in previous cycles that
includes; Standalone roles, nftables switch, OVN RAFT support, ephemeral
heat for upgrade use cases, CI jobs for Deployed Ceph and Ceph
Container Promotion in CI etc.
There was a
discussion on the suggestion to stop backporting feature removals from
stable branches as we did for wallaby in the last few cycles. Though
TripleO does not follow stable policy, backporting feature removals with
dependent patches across repos more often than not leads to broken CI
and user support cases. We would be well placed if we can identify and
do those before zed release and avoid the firefighting later.
[Action Items]
- Update zed review priorities etherpad[1] with deprecation/removal patches [All]
# Standalone Ansible Roles Status Update
James
provided an update on the current status of "Standalone Ansible Roles'
POC and we discussed the possible future scope of it beyond the current
"scale-out compute" use case. He also demoed molecule based automation
of deploying/testing a compute node with the new meta-role. Chandan
shared the current work on CI jobs/scenarios to deploy computes with
standalone roles.
[Action Items]
- Prioritize review/merge of proposed patches[2] on this topic [tripleo-core]
- Finish the CI jobs for testing standalone roles and make them voting [Chandan/tripleo-ci-core]
# Standalone Ansible Roles and External Ceph
John
presented his work to integrate computes deployed with standalone roles
with external ceph. We don't plan to extend this effort to include
tripleo deploying ceph using standalone roles, unless there is a
requirement for that in the future.
[Action Items]
- Prioritize review of ceph related standalone role patches.[tripleo-core]
# Multi-rhel support for Compute
Brendan
gave an update on the ongoing effort for multi-rhel support, different
options evaluated and current POC status with composable roles approach.
There is still some work left on automating role split during upgrade,
which is expected to be completed this cycle.
[Action Items]
- Finish composable roles POC and test upgrade procedure manually [upgrades team]
- Add CI jobs to test the proposed procedure upstream as feasible [upgrades team]
- Implement role-split automation during upgrade [Takashi/Rabi]
# OpenStack SDK, Ansible Collections OpenStack and TripleO
We
discussed the issues around openstacksdk backward incompatibility and
the current pinning for it in RDO/TripleO. We agreed on an action plan
to pre-release ansible-openstack-collection (AOC) for RDO/TripleO
(though it is still not fully compatible with latest openstacksdk) and
remove the pin so that Zed RDO can be released with openstacksdk
0.101.0.
[Action Items]
- Bump openstacksdk to 0.101.0 for RDO master and zed [Jakob/Alfredo]
- Update RDO ansible-collections-openstack
to the 2.0.0 pre-release [Jakob/Alfredo]
# TripleO and Distributed Project Leadership (DPL)
In
this session, we discussed the possibility of opting for DPL in TripleO
and sharing the responsibilities across multiple leaders in the team.
There were a number of volunteers for different liaison roles and we in
general agreed that it would be a good thing to do. We would propose the
required change to governance sometime this cycle, unless there is any
concern.
[Action Items]
- Finalize all required liaison roles and propose patch to openstack governance repo [Rabi]
# Migration from puppetlabs/apache
As
part of the broader effort to reduce the puppet footprint, Cedric
presented the ongoing effort to replace puppetlabs/apache with the new
ansible role "tripleo_httpd_vhost"[3].
A few services have already been migrated and remaining ones are
expected to be done this cycle with broader participation from sub
teams.
[Action Items]
- Review/merge the already proposed patches for migration [tripleo-core]
- Plan and work on migrating remaining services [Cedric]
# State of CI
CI team presented on a vast range of topics that included,
- Added coverage of upgrade jobs across upstream check/gate, periodic
component pipeline and periodic integration pipeline in zed (Marios)
- Upstream jobs for multi-rhel testing (Marios)
- Current state of OVB Jobs (Chandan)
- TripleO CI Jobs on IBM Cloud and the new feature to hold nodes for troubleshooting (Chandan)
- Tempest Allow List (Arx/Pooja/Soniya)
- Tempest Dashboard (Arx/Lukas)
# Container Capabilities
We
revisited the topic of privileged containers for services and how we
can possibly avoid those by allowing the required limited container
capabilities. As all these services are mostly for the compute role, we
agreed that it would probably be better to align this effort with
"Standalone Roles" for better testing.
# OS Migrate Status Update
Jirka
provided an update on the tenant migration tool 'os-migrate', an
ansible collection that is not a TripleO umbrella project, but used by
the upgrades team.
It's possible that I missed
some important points from these discussions. Please feel free to add
them by replying to this thread. All session etherpad links are in the
main schedule etherpad[4] for reference. I'll share the session recordings once they're made
available to us.
Thanks Again.
--