#opnfv-meeting: Weekly Technical Discussion
Meeting started by bh526r at 14:01:05 UTC
(full logs).
Meeting summary
- Roll Call (bh526r, 14:01:14)
- Bin Hu (bh526r,
14:01:19)
- Al Morton (bh526r,
14:01:25)
- Georg Kunz (bh526r,
14:01:33)
- Juha Oravanien (bh526r,
14:01:42)
- Prakash Ramchandran (bh526r,
14:01:51)
- Uli Kleber (uli-k,
14:02:02)
- Dan Radez (bh526r,
14:02:04)
- Aimee Ukasick (bh526r,
14:03:19)
- Bryan Sullivan (bh526r,
14:03:25)
- rprakash (rprakash,
14:05:47)
- Proposal of DevOps Mode; (bh526r, 14:07:10)
- One of the topic is Keystone integration
(bh526r,
14:07:31)
- If I need to have an OPNFV build, should I be
able to pull from stable branch or from master with more recent
feature (bh526r,
14:08:08)
- rprakash (rprakash_,
14:08:28)
- This assumption needs to be validated
(bh526r,
14:08:40)
- (Uli) we can try it to validate the
feasibility (bh526r,
14:09:28)
- (Bryan) another implication is, e.g. if I have
a test, but dovetail cannot pick it up for a few months, how stable
will be the test and python library while we are moving
forward (bh526r,
14:10:47)
- So we don't leave behind the community too far,
while we don't have to wait just for the sake of stable
release (bh526r,
14:11:32)
- We need to really understand who the users
are (bh526r,
14:11:46)
- (Uli) we have 2 types of users (bh526r,
14:11:54)
- One type is to use the coolest features
(bh526r,
14:12:16)
- The other type is to use the newest
features (bh526r,
14:12:52)
- Larry Lamers (bh526r,
14:13:32)
- Tapio (bh526r,
14:13:35)
- We need to measure the user community
proactively and understand their needs (bh526r,
14:14:14)
- Talking about the first type of users, some are
still trying to use Brahmaputra, even if we don't support it any
more (bh526r,
14:14:58)
- We need to test feasibility of deploying OPNFV
scenarios on trunk? Which installers support that? (bryan_att,
14:15:39)
- (Bryan) I am looking for a stable development
environment with stable features that enable me to move
forward (bh526r,
14:17:50)
- ability to devlopoment on trunk on Openstack
like with Congress (rprakash_,
14:21:47)
- (rprakash) ability to development on trunk with
OpenStack Murano, Manila etc. (rprakash_,
14:23:24)
- (aimeeu) using Quickstart with trunklevel
developement and Dan says in Newtorn Tacker is in (rprakash_,
14:24:10)
- (Bryan) ability to develop with multiple
components from different upstreams to work together from trunk to
suport new features (rprakash_,
14:26:26)
- (Bryan) Dvelopment envrionment like RDO,
triple-O, Packstack, Quickstart and understand which one are
suitable for different features and test plus continuous Devlopment
to support OPNFV community (rprakash_,
14:28:48)
- (bryan) Feature Devlopment role is different
from Packaging and Delivering different upstreams to the
releases (rprakash_,
14:30:23)
- (Bryan) Resilliancy is imporant for getting
Control Plane stability and needs simplicty and flexibility of
control plane (rprakash_,
14:31:41)
- (Bryant) ability add drop Cielometer for say
Billing , Add drop Congree like Policy usgae is different etc. So
customization of target deployment is one aspect (rprakash_,
14:33:07)
- (Bryan) How light the Control Plane can be with
flexibility of installers to choose some of the options from
OpenStack for specialized environment that Service Provider
needs (rprakash_,
14:34:31)
- Uli will try to put those wishes into scenario
consolidation (bh526r,
14:35:25)
- scenario definitions (bh526r,
14:36:02)
- (Bryan) Orchestrator like JujU, Puppet have
ability to slect OpenStack Modules and establish relationships and
if Scenarios can configure similarly that will be the goal
(rprakash_,
14:38:32)
- look at scenario descriptor file and pod
descriptor file (bh526r,
14:40:27)
- (uli) ability to orchetsrate in layers for
Hardware(POD), Infra(descriptor), VNFD/NSD (MANO descriptor)
etc (rprakash_,
14:42:34)
- (Bryan) mentions like L4 service depends of L3
, L2, so there may be cross layer dependency (rprakash_,
14:43:13)
- (Bryan) Its possible to add and remove Charm
using Juju and similar can be done with Puppet, Ansible playbook
etc. (rprakash_,
14:44:11)
- (Bryan) Ability to add reomove Tacke eg. could
be using VNF using Tacker once and next using JuJu
dynamically (rprakash_,
14:46:28)
- (Dan Radez) Why do we need this is there value
to this? (rprakash_,
14:47:08)
- (Bin) How we can leverage Scenarios to do
flexible Control plane objectives and lets discuss and come up with
some solutions (rprakash_,
14:48:18)
- Update of Scenario Consolidation (bh526r, 14:49:12)
- (Uli) has submitted a patch in Octopus to be
reviewed- request for reveiew (rprakash_,
14:49:32)
- Uli submitted a patch of skeleton of scenario
document set (bh526r,
14:49:32)
- https://gerrit.opnfv.org/gerrit/#/c/28443
(uli-k,
14:49:59)
- Uli will send an email to mailing list so that
everyone can review the patch (bh526r,
14:51:37)
- Then we can move forward to review scenario
descriptor patches (bh526r,
14:52:04)
- to summarize what we will be doing toward the
last topic areas (1) assessing developer-focused options for
OpenStack, ODL test envs simpler than full OPNFV deploys, and
providing advice on the wiki. (2) assessing over time whether that
gives us more dev freedom at a good balance wrt re platform
stability (bryan_att,
14:53:24)
- (Bryan) mentioned that there was good
discussion in MANO WG. (bh526r,
14:54:55)
- MANO Update (bh526r, 14:55:06)
- (3) looking at how flexible the installers can
be for customizing the deployment (e.g. thru the descriptor files);
(4) how modeling concepts can be applied to the control plane as
well, e.g. how well the blueprint model can cover the needs of the
various descriptors we use in CI/CD (bryan_att,
14:55:11)
- the goal of (4) being to someday have a common
abstraction/specification for the intended control plane deployment,
that the installers can "onboard" and then take action to
deploy; (bryan_att,
14:56:18)
- David M is not present, so Milestone Exception
Process discussion deferred to next time (March 2) (bh526r,
14:58:01)
- Cross-Community CI process will be on March
2 (bh526r,
14:59:34)
- and E release plan (bh526r,
14:59:45)
- (rprakash) MANO release E has learining from
vIMS VNF about different approach to orchestartion through
installers and without it, plus testing (rprakash_,
15:00:01)
- (rprkash) the other aspect realted to keystone
v 3 v 2 usage based on installers support (rprakash_,
15:00:55)
- (rprakash) the bestpractices is still under
review for VNF packaing, onboarding, LCM etc. (rprakash_,
15:01:53)
- Next week (Feb 23), we will be dedicated to
discuss 2 new project proposals (bh526r,
15:02:53)
- (rprakash) About the Architecture tehre are
different views from different work groups like pipeling of
CI,CT,CV, CD form iintegration, Test, Vlaidation to Deployment and
Tapio has a Software Architecture viewpoint and thus we need to
reflect the upstream integration aspect as we are looking now at
Bifrost and working through up[straem Trunk in CI (rprakash_,
15:04:06)
- On March 2, we will discuss (1)
DevOps/Continuous Delivery and How Scenario Descriptor can
enable/support it (2) Cross-Community CI Process (3) Milestone
Exception Process (4) E Release Plan (bh526r,
15:04:27)
- And the perhaps (5) MANO WG Update and MANO
Architecture if time permits (bh526r,
15:04:55)
- Meeting adjourned (bh526r,
15:05:09)
Meeting ended at 15:06:52 UTC
(full logs).
Action items
- (none)
People present (lines said)
- bh526r (45)
- rprakash_ (23)
- bryan_att (5)
- collabot` (3)
- rprakash (2)
- uli-k (2)
- radez (1)
Generated by MeetBot 0.1.4.