#opnfv-meeting: Weekly Technical Discussion
Meeting started by bh526r at 14:04:36 UTC
(full logs).
Meeting summary
- Roll Call (bh526r, 14:06:54)
- Bin Hu (bh526r,
14:06:59)
- Tim Irnich (timirnich,
14:07:09)
- Bryan Sullivan (bh526r,
14:07:28)
- Claudio Serra (bh526r,
14:07:36)
- Dave Neary (bh526r,
14:07:44)
- Pierre Lynch (bh526r,
14:07:51)
- Tapio Tallgren (bh526r,
14:08:41)
- Dan Radez (bh526r,
14:08:57)
- Deep Dive of Release Goals of Model and VES Projects in Fraser Release (bh526r, 14:09:31)
- Background: (bh526r,
14:09:42)
- https://lists.opnfv.org/pipermail/opnfv-tech-discuss/2017-November/019254.html
(bh526r,
14:09:57)
- Bryan reviewed briefly the background and
evolving of OPNFV with the new era of ONAP emerging on market
(bh526r,
14:10:51)
- Tim asked if we do integration and testing in
OPNFV, will we end up with still developing some components that are
not in ONAP? Is it still the same situation as of today?
(bh526r,
14:13:36)
- Bryan mentioned that we may not use
installer-based approach, but more focus on container, k8s,
etc. (bh526r,
14:14:42)
- So it is more like VIM-agnostic approach
(bh526r,
14:15:06)
- e.g. Model experiment now is not dependent on
OpenStack (bh526r,
14:15:22)
- Narinder Gupta (bh526r,
14:15:41)
- In a word, stop relying on installers
(bh526r,
14:17:37)
- Then what should OPNFV do for ONAP? For
example, testing, certification, more complex use case such as
hybrid cloud (bh526r,
14:18:16)
- Tim understands the frustration in the past,
e.g. pain of installers. OPNFV is not expected to be
production-ready. (bh526r,
14:22:12)
- Happy to see the path of evolution, e.g.
evolution to more VIMs such as k8s (bh526r,
14:22:39)
- Dave Neary indicates that we still need
something to manage k8s instead, so not sure if simplification can
be achieved even if we evolve to k8s. Installer just provides a way
to install and deploy. (bh526r,
14:25:22)
- Tim understands the complexity is not expected,
and wants some level of simplification. (bh526r,
14:27:20)
- The question is whether to focus on
simplification, or to focus on support of more VIMs (bh526r,
14:28:34)
- Bryan indicated less dependencies, and some
more internal collaboration (bh526r,
14:29:23)
- Tim thinks it is very important topic, and if
VES/Models finds a way to simplify the experience, we OPNFV needs to
look into it (bh526r,
14:30:37)
- Dan mentions that maybe we need to see if this
issue is shared by broader community, or is specific to the practice
that one specific project needs. Then we can address it at broader
community level or support specific project(s) (bh526r,
14:32:16)
- Tapio understands because Nokia experienced the
same when using different installers. For example, if anything goes
wrong, we need to go back to the beginning of re-install which takes
another 2 hours (bh526r,
14:33:49)
- So it would be good to separate into different
layers/modules (bh526r,
14:34:15)
- Tim thinks it is ideal to put it on the agenda
of PlugFest because of the complexity of this topic, i.e. (1)
supporting multiple VIMs/Clouds in integration instead of
OpenStack-centric (2) Improve user experience. (bh526r,
14:39:33)
- Bryan mentions that it is really not to give
actions to installers, but to reduce the dependency of feature
projects on installers (bh526r,
14:41:20)
- ACTION: Tim will
arrange a session in PlugFest for this topic. Bryan will help scope
this right, and give the right objective (bh526r,
14:42:40)
- Discuss Collected Feedback on Hardware Requirements and Improvements on Pharos Specification (bh526r, 14:44:32)
- Background: (bh526r,
14:44:39)
- https://wiki.opnfv.org/display/pharos/Pharos+Specification
(bh526r,
14:44:59)
- https://etherpad.opnfv.org/p/Pharos_Specification2.0
(bh526r,
14:45:24)
- Tim mentions that there has been discussions of
integration of L2 gateway of OpenStack, and more testings in HA
projects. (bh526r,
14:46:27)
- SmartNIC is another example (bh526r,
14:47:21)
- We try to figure out a way of how to deal with
it, and avoid situation that certain thing only works in certain
lab (bh526r,
14:48:01)
- One element of this is Pharos spec (bh526r,
14:48:14)
- Others include CI engine (intelligent to
differentiate labs), test framework (intelligent to activate or
deactivate test cases automatically) (bh526r,
14:49:41)
- Bryan mentions "programmable lab", i.e. users
can use interface for lab to present some production features
(bh526r,
14:54:36)
- Tim indicates that today's Pharos Spec
describes server features (bh526r,
14:56:42)
- Tim will advocates it in several places such as
infra, testing, etc. (bh526r,
14:58:00)
- Bin will follow up with Jack Morgan regarding
Phraos Spec 2.0 (bh526r,
14:58:21)
- AOB (bh526r, 14:59:44)
- Nothing else, meeting adjourned (bh526r,
14:59:53)
Meeting ended at 14:59:57 UTC
(full logs).
Action items
- Tim will arrange a session in PlugFest for this topic. Bryan will help scope this right, and give the right objective
People present (lines said)
- bh526r (49)
- collabot (3)
- timirnich (1)
Generated by MeetBot 0.1.4.