#opnfv-meeting: Weekly Technical Discussion
Meeting started by bh526r at 12:58:55 UTC
(full logs).
Meeting summary
- Roll Call (bh526r, 12:59:05)
- Bin Hu (bh526r,
12:59:11)
- Cedric Ollivier (bh526r,
12:59:47)
- Mark Shostak (bh526r,
13:00:00)
- Greg Oberfield (bh526r,
13:00:11)
- Sridhar Rao (bh526r,
13:00:20)
- Matt McEuen (bh526r,
13:00:43)
- Trevor Cooper (bh526r,
13:00:52)
- Cédric Ollivier (ollivier,
13:02:09)
- Al Morton (bh526r,
13:02:32)
- Georg Kunz (bh526r,
13:02:39)
- Continue to Discuss New Project Proposal "Airship-based Infrastructure Deployment and Lifecycle Management" (bh526r, 13:04:08)
- https://wiki.opnfv.org/display/PROJ/Project+Proposals+Airship
(bh526r,
13:04:29)
- Trevor Cooper (trevor_intel,
13:04:54)
- Greg Oberfield from AT&T gave an update of
resolved comments after communicating with stakeholders (bh526r,
13:05:15)
- Two major comments/comments are
addressed (bh526r,
13:05:36)
- First major comment/concern: "Why multiple
Installers and how to address fragmentation issues in the
past?" (bh526r,
13:06:42)
- Greg and team communicated community, and
explained that the root cause of prior fragmentation issue is
lacking a standard definition of NFVi (bh526r,
13:10:06)
- Mark Beierl (bh526r,
13:10:19)
- Cedric full agrees that multiple installers is
not an issue at all (bh526r,
13:10:45)
- The main issue is lacking common definition,
and loses lots of end users (bh526r,
13:11:20)
- Daniel Balsiger (bh526r,
13:11:43)
- Pierre Lynch (bh526r,
13:12:08)
- Parth (bh526r,
13:12:16)
- Cedric asked why Airship cannot be merged with
XCI, and MaaS (bh526r,
13:13:02)
- #info Matt explained that Airship 1.0 used MaaS
for bare metal provisioning, which was the best tool at that
time (bh526r,
13:13:56)
- With Airship 2.0 work in progress, we are going
to leverage k8s Cluster API for bare metal provisioning (bh526r,
13:14:50)
- i.e. metal3 project, and lots of plugins for
AWS, Google cloud etc. (bh526r,
13:15:20)
- The 1st driver to leverage is Ironic
(bh526r,
13:15:38)
- And Kubeadm support as tooling for k8s
itself (bh526r,
13:16:00)
- We are behind those projects for k8s cluster
for bare metal provisioning (bh526r,
13:16:26)
- Lots of installers there, e.g.
Ansible/Kubespray/Bifrost in XCI, and Airship isn't intended to
support other installers, which is out of scope of mission of
Airship (bh526r,
13:17:36)
- Each installer can focus on its good, and
particular use cases (bh526r,
13:17:56)
- Cedric agreed that we shouldn't merge them
together (bh526r,
13:18:22)
- Greg agreed to add this point in presentation
why XCI and Airship are not appropriate to merge together
(bh526r,
13:19:07)
- Continue to capture the resolved comments from
Greg's update slides: (bh526r,
13:23:42)
- Re: multiple installers: multiple installers
bring more choices for end users, and encourages technological
innovation (bh526r,
13:24:16)
- With standard definition of common NFVi, all
installers will implement it with the same result. Thus
fragmentation issue goes away. And end users will have more choices
when shopping on market (bh526r,
13:24:45)
- Team will ensure level playground of all
technologies / installers – no privilege for any installer or
technology (bh526r,
13:25:16)
- 2nd Resolved comments "How to add new features
in Airship, e.g. DPDK, and bring new resources?" (bh526r,
13:25:49)
- All new feature work in Airship is happening in
OpenStack: (bh526r,
13:26:19)
- https://airship.atlassian.net/projects/AIR/issues
(bh526r,
13:26:54)
- DPDK, Tungsten Fabric etc. is happening in OSH,
and brought to Airship on June 18 to bring new NFV features
(upstream) (bh526r,
13:27:16)
- Features are market-driven. Business will
invest resources if they see ROI, i.e. benefit end users.
(bh526r,
13:27:33)
- Airship 2.0 is modular design, e.g. something
for host provisioning + Airship for workloads on top of
cluster (bh526r,
13:27:52)
- Because Manuel is on vacation and absent from
discussion today but he made comments last week, Bin indicated that
team communicated with Manuel offline and resolved his comments and
concerns. Manuel agreed to move forward and bring Airship proposal
for TSC creation review and approval. (bh526r,
13:29:48)
- Because no further comments, and all comments
and concerns are resolved, group reaches consensus and agrees to
recommend "Airship-based Infrastructure Deployment and Lifecycle
Management" project proposal for TSC creation review and
approval. (bh526r,
13:31:30)
- AOB (bh526r, 13:32:15)
- Bin asked everyone to propose items for
discussion for future meetings (bh526r,
13:32:35)
- Meeting adjourned (bh526r,
13:32:45)
Meeting ended at 13:33:55 UTC
(full logs).
Action items
- (none)
People present (lines said)
- bh526r (50)
- collabot` (3)
- acmacm (1)
- trevor_intel (1)
- ollivier (1)
Generated by MeetBot 0.1.4.