#opendaylight-meeting: tsc
Meeting started by colindixon at 17:00:30 UTC
(full logs).
Meeting summary
- agenda bashing and roll call (colindixon, 17:00:41)
- regXboi (thinks he's edwarnicke) (regXboi,
17:00:46)
- https://meetings.opendaylight.org/opendaylight-meeting/2015/tsc/opendaylight-meeting-tsc.2015-07-09-17.00.html
Minutes from last week’s meeting (tbachman,
17:00:48)
- dfarrell07 for cdub/Red Hat (dfarrell07,
17:00:53)
- https://wiki.opendaylight.org/index.php?title=TSC:Main&oldid=33638#Agenda
the agenda in it’s usual place (colindixon,
17:01:01)
- dlenrow (dlenrow,
17:01:05)
- colindixon (colindixon,
17:01:12)
- abhijitkumbhare for Chris Price /
Ericsson (abhijitkumbhare,
17:01:18)
- ACTION: colindixon to
try to find somebody to help with documenting the general procedure
for the platform upgrade from Helium to Lithium (for SR1)
(colindixon,
17:01:20)
- ACTION: ChrisPriceAB
to work with rovarga, jmedved, et. al. to look into leveraging OPNFV
infrastructure for performance measurements and (colindixon,
17:01:25)
- mohnish anumala (mohnish,
17:01:38)
- LuisGomez (LuisGomez,
17:02:18)
- ACTION: gzhao to send
out exact cutoff dates/times for Lithium SRs and get him to if he
hasn’t (colindixon,
17:02:58)
- ACTION: phrobb and
gzhao to beat the bushes to try to get cherry-picks out for
Helium-SR4 (colindixon,
17:03:33)
- ACTION: phrobb to
send out mail when the new infra person is up-to-speed so that he
knows when he will not be waking people up after hours (colindixon,
17:03:57)
- colindixon and dfarrell07 talked about past
#action to notify projects of test failures. We already do that via
the Jenkins mailing list and topics. Documented it better in project
getting started wiki. (dfarrell07,
17:04:17)
- ACTION: tony and
zxiiro to look at the sonar/jacoco reports and try to figure out how
(and if) we can reasonby get feature-level code coverage
information (colindixon,
17:04:21)
- snackewm (for Uri) (snackewm,
17:04:25)
- ACTION: tony to send
out an e-mail explaining his alternative solution to version bumping
and branch cutting (colindixon,
17:04:46)
- ACTION: regXboi to
send out email on continuous release process (colindixon,
17:04:49)
- events (colindixon, 17:06:33)
- http://www.opendaylight.org/news/events/
ODL Events page (tbachman,
17:06:39)
- CFP for OpenStack closed yesterday, hopefully
people sent things in (colindixon,
17:06:56)
- https://wiki.opendaylight.org/view/Events:Be_Dev_Forum
pleas add developer design topics here (colindixon,
17:07:12)
- we need to be building the agenda for the dev
forum in the next week, so sooner rather than later would be
good (colindixon,
17:07:38)
- IETF hackathon is on the 18th and 19th (this
week). Contact Charles Eckel (eckelcu@cisco.com) for more
information (tbachman,
17:08:16)
- stable/helium and stable/lithium updates (colindixon, 17:08:24)
- gzhao says that for stable/helium, we just need
a date (colindixon,
17:08:36)
- there have been some updates, but not a
lot (colindixon,
17:08:47)
- current currently schedule for 7/23 vote, which
means cut on Sunday at 11:59p UTC (colindixon,
17:09:10)
- colindixon asks if folks need more time to ship
SR4, beyond the additional week already allocated (tbachman,
17:09:40)
- abhijitkumbhare asks if it makes sense to ship
Helium SR4 at the same time as the Lithium SR1 release (tbachman,
17:10:06)
- gzhao says OPNFV has released ARNO — asks when
their first stable release occurs, which may be important, given
that ODL is their upstream (tbachman,
17:10:33)
- dfarrell07 says he doesn’t believe there’s a
stable release planned (tbachman,
17:10:48)
- colindixon says his gut reaction is to ship SR4
with the current 1-week delay, unless he hears otherwise
(tbachman,
17:11:46)
- regXboi asks how long have we been holding
SR4? (tbachman,
17:12:06)
- colindixon says a long time — we’ve only
slipped it once (tbachman,
17:12:18)
- regXboi asks how long have we gone through last
call to wait for folks to make last call — has it been more than 2
weeks? (tbachman,
17:13:03)
- colindixon says yes (tbachman,
17:13:06)
- https://lists.opendaylight.org/pipermail/release/2015-July/003076.html
(colindixon,
17:13:18)
- regXboi asks have we been in "last call"
internally for more than two weeks (regXboi,
17:13:33)
- https://lists.opendaylight.org/pipermail/release/2015-July/003076.html
colindixon says we’ve been doing last call for SR4 since July 1st
(colindixon,
17:13:48)
- regXboi says ship it (regXboi,
17:14:09)
- dfarrell07 says ship it (dfarrell07,
17:14:25)
- zxiiro asks if this is the last SR for
stable/helium (tbachman,
17:14:45)
- colindixon says yes (tbachman,
17:14:48)
- zxiiro says the reason for his question is
because he’s wondering if we need to do version bumps post
release (tbachman,
17:15:17)
- colindixon says yes (tbachman,
17:15:17)
- zxiiro says he’s on PTO next thursday, but
could start that before he leaves (tbachman,
17:15:30)
- https://lists.opendaylight.org/pipermail/release/2015-July/003138.html
this is the start of the discussion for Lithium-SR1 timing (colindixon,
17:19:23)
- it sounds like we might pus Lithium-SR1 back by
a week to allow for more time after the summit (colindixon,
17:19:39)
- we’ll cut the artifacts a full week in advance
at 11:59p UTC to allow for time for testing (colindixon,
17:20:01)
- regXboi says a project announces in M3 whether
a feature is stable, but *when* does the TSC do the review?
(regXboi,
17:23:02)
- it’s evaluated by the TSC during the release
review (colindixon,
17:23:19)
- that means there needs to be a quorum of the
TSC for those projects (regXboi,
17:23:52)
- to be clear, a project must be mature to have
stable feautres, not the other way around (colindixon,
17:27:06)
- dfarrell07 asks how often and when are we
reviewing stable feautres to be stabel again (colindixon,
17:27:42)
- colindixon answers that stable features will be
reviewed per-release (dfarrell07,
17:28:35)
- stable features are only defined in the context
of Be at teh meoment (colindixon,
17:28:37)
- if we copy it moving forward, we will re-review
stable stable features per-release (colindixon,
17:29:00)
- LuisGomez asks if a feature could move to beign
stable in an SR even if they aren’t stable in the main
release (colindixon,
17:29:31)
- colindixon says his gut reaction is no, but
he’s open to other arguments (colindixon,
17:29:48)
- LuisGomez sees no reason not to allow
(colindixon,
17:29:54)
- phrobb says from his standpoint, the question
is back to the integration team — if we allow a feature to become
stable, as it changes where it is in the karaf distribution
(tbachman,
17:30:56)
- abhijitkumbhare asks for clarification how
feature maturity affects karaf distribution (tbachman,
17:32:05)
- colindixon says there will be two karaf feature
repositories — one for stable features, and another for non-stable
features (tbachman,
17:32:33)
- colindixon says there will also be two
distributions — one that is just stable features, and another that
includes nono-stable features (tbachman,
17:32:55)
- LuisGomez says a stable feature is not declared
until the end of a release, which makes it tough to include in the
release (tbachman,
17:33:42)
- phrobb notes the changes would need to occur
in our karaf distributions and their testing well as
documentation (tbachman,
17:34:08)
- colindixon says an advantage of stable features
means a stable distribution which (in theory) means that it’s less
likely to break (tbachman,
17:34:33)
- regXboi asks a bigger question - is the Be plan
being voted on as written, or will the dates be adjusted by 2
weeks? (regXboi,
17:34:59)
- did you just chop M1 time by 2 weeks?
(regXboi,
17:35:35)
- colindixon says there are 3 +1 votes to the
release plan on the mailing list (tbachman,
17:35:39)
- I will now channel edwarnicke and say "you
can't do that" (regXboi,
17:36:11)
- colindixon says that the release plan has been
available for some time (tbachman,
17:36:20)
- regXboi says he is channelling edwarnicke
:) (regXboi,
17:36:52)
- VOTE: Voted on "Should
the TSC approve the Beryllium Release Plan, as per this email:
https://lists.opendaylight.org/pipermail/tsc/2015-July/003453.html
?" Results are, +1: 7, -1: 1 (tbachman,
17:39:39)
- AGREED: The TSC
approves the Beryllium Release Plan, as per this email:
https://lists.opendaylight.org/pipermail/tsc/2015-July/003453.html
(tbachman,
17:39:50)
- regXboi's vote is based on his understanding of
edwarnicke's opinions about dates (regXboi,
17:40:48)
- Fabric as a Service Creation Review (tbachman, 17:41:31)
- https://wiki.opendaylight.org/view/Project_Proposals:FaaS
project proposal (colindixon,
17:41:40)
- https://lists.opendaylight.org/pipermail/project-proposals/2015-June/000323.html
proposed on 6/29/2015 (colindixon,
17:42:03)
- ACTION: xingjun will
post the slides on the wiki project proposal after the review
(colindixon,
17:42:50)
- problem statement is that is that nthe network
is falling bheind int terms of capex/opex reduction, agaility and
vendor lock-in (colindixon,
17:43:28)
- The problem statement is that the computing
paradigm is evolving inside and outside the Data Center: CAPEX/OPEX
reduction; Service agility/Automation/Efficiency; No vendor
lock-in (tbachman,
17:43:40)
- The network is falling behind (tbachman,
17:43:46)
- Network Service Operation Today: using
low-level tools/APIs/interfaces (e.g. CLI); heavily depends on
manual work, wtih no service agility (tbachman,
17:44:30)
- Industry Initiatives/Solutions: SDN-1 (Control
and Data Plane separation) — protocol standardization; SDN-2
(APplication Centric Modeling) GBP/NIC/NEMO — top-down approaches,
monolithic solution and coupled with low-level primitives, making it
hard to implement on 3rd psrty devices (tbachman,
17:46:40)
- multi layer abstraction is required, no one
layer fits all (tbachman,
17:46:53)
- Fabric as a Service is a bottom-up abstraction
— keep network primitives familiar to CT personell; defines FaaS
model — unified service via Fabric (tbachman,
17:49:54)
- Fabric is a set of network resources (usually
network node and topology between them) within the same control
plane (tbachman,
17:50:21)
- A fabric is a self-managed unit (tbachman,
17:50:29)
- a fabric has a logical centralized management
interface/control logic/state (tbachman,
17:50:40)
- a fabric provides common network services
(connectivity — L2/L3 logical network element: logical switch,
logical router, gateway, tunnel end points) (tbachman,
17:51:22)
- a Fabric manager manges fabric — CRUD
operations of fabric objects (tbachman,
17:51:39)
- Orchestrate/coordinate services across multiple
fabrics (tbachman,
17:52:01)
- resource manaagement occurs across
fabric (tbachman,
17:52:11)
- The fabric manager interacts with other
dependent ODl modules (tbachman,
17:52:28)
- The deliverables are are FaaS yang models and a
Fabric Manager module (tbachman,
17:53:16)
- The FaaS has an existing code base — UI, model,
code for VLAN, and Fabric Manager (tbachman,
17:55:00)
- LuisGomez asks which protocol and devices are
going to be targeted first — or is that specified in the project
poroposal (e.g openflow, OVS, etc.) (colindixon,
17:56:09)
- xingjun says they’re going to leverage existing
SB plugins in ODL — OVSDB and openflow to support openflow
switches (tbachman,
17:56:17)
- LuisGomez says in the scope of this release
those two implementations should be available (tbachman,
17:56:33)
- colindixon says there are 3 existing projects
that do this: VTN , OVSDB net-virt, and GBP. Why is a new project
needed rather than leveraging existing ones (tbachman,
17:57:10)
- xingjun says they don’t overlap (tbachman,
17:57:23)
- colindixon says there might be differences, but
they definitely overlap (tbachman,
17:57:32)
- one question is how would they co-exist?
(regXboi,
17:57:49)
- xingjun says they complement each other — one
layer can’t solve all the problems. (tbachman,
17:58:00)
- dlenrow says that’s why the other projects
exist — the map an abstraction onto the SB protocols (tbachman,
17:58:15)
- colindixon notes that he was asking for his
edification, historically, the TSC does not mandate that incubation
projects have nonoverlapping scope (colindixon,
17:58:57)
- regXboi asks in the Neutron example, neutron
puts the information in the MD-SAL and make it available to all the
conusming projects; how would FaaS coexist with GBP or
OVSDB-net-virt or VTN? (tbachman,
17:59:23)
- xingjun says you could use GBP to map the model
as a service if you want (tbachman,
18:01:05)
- dlenrow asks if all the existing projects (GBP,
OVSDB-net-virt, VTN) would use this new FaaS? (tbachman,
18:01:37)
- xingjun yes, that’s a possibility (tbachman,
18:01:42)
- dlenrow says the NIC project plans to also
build this functionality, and doesn’t plan to do it in a monolithic
fashion, and don’t see the need for FaaS (tbachman,
18:02:40)
- colindixon says there are some TSC members who
would like to see more discussion on the project over the mailing
list before voting (tbachman,
18:03:08)
- regXboi says we need to be sure that this
project has some way of coexsiting with other projects, and isn’t
seeing it just yet (tbachman,
18:03:36)
- LuisGomez says we already have this issue in
ODL (tbachman,
18:03:47)
- ACTION: colindixon to
start a thead on the mailing list followed by a vote eithe ron the
mailing list, followed by a vote on the mailing list and/or next
week’s meeting (colindixon,
18:03:53)
Meeting ended at 18:05:15 UTC
(full logs).
Action items
- colindixon to try to find somebody to help with documenting the general procedure for the platform upgrade from Helium to Lithium (for SR1)
- ChrisPriceAB to work with rovarga, jmedved, et. al. to look into leveraging OPNFV infrastructure for performance measurements and
- gzhao to send out exact cutoff dates/times for Lithium SRs and get him to if he hasn’t
- phrobb and gzhao to beat the bushes to try to get cherry-picks out for Helium-SR4
- phrobb to send out mail when the new infra person is up-to-speed so that he knows when he will not be waking people up after hours
- tony and zxiiro to look at the sonar/jacoco reports and try to figure out how (and if) we can reasonby get feature-level code coverage information
- tony to send out an e-mail explaining his alternative solution to version bumping and branch cutting
- regXboi to send out email on continuous release process
- xingjun will post the slides on the wiki project proposal after the review
- colindixon to start a thead on the mailing list followed by a vote eithe ron the mailing list, followed by a vote on the mailing list and/or next week’s meeting
Action items, by person
- colindixon
- colindixon to try to find somebody to help with documenting the general procedure for the platform upgrade from Helium to Lithium (for SR1)
- colindixon to start a thead on the mailing list followed by a vote eithe ron the mailing list, followed by a vote on the mailing list and/or next week’s meeting
- gzhao
- gzhao to send out exact cutoff dates/times for Lithium SRs and get him to if he hasn’t
- phrobb and gzhao to beat the bushes to try to get cherry-picks out for Helium-SR4
- phrobb
- phrobb and gzhao to beat the bushes to try to get cherry-picks out for Helium-SR4
- phrobb to send out mail when the new infra person is up-to-speed so that he knows when he will not be waking people up after hours
- regXboi
- regXboi to send out email on continuous release process
- UNASSIGNED
- ChrisPriceAB to work with rovarga, jmedved, et. al. to look into leveraging OPNFV infrastructure for performance measurements and
- tony and zxiiro to look at the sonar/jacoco reports and try to figure out how (and if) we can reasonby get feature-level code coverage information
- tony to send out an e-mail explaining his alternative solution to version bumping and branch cutting
- xingjun will post the slides on the wiki project proposal after the review
People present (lines said)
- tbachman (106)
- colindixon (81)
- regXboi (41)
- odl_meetbot (17)
- dfarrell07 (13)
- abhijitkumbhare (8)
- gzhao (6)
- ebrjohn (3)
- mohnish (3)
- dlenrow (2)
- icbts (2)
- phrobb (2)
- LuisGomez (2)
- hideyuki (2)
- snackewm (2)
Generated by MeetBot 0.1.4.