►
From YouTube: Kubernetes SIG Architecture 20180913
Description
A
Welcome
everybody:
it
is
Thursday
September,
13
2018,
it
is
the
kubernetes
sig
architecture,
meeting
I'm
your
co-chair
Jay
singer
Dumars
and
the
other
chair
is
Brian
grant
who
will
be
joining
us
momentarily
and
we're
going
to
be
going
through.
The
agenda,
which
is
located
at
Bentley,
slash
sake,
architecture
and
first
thing
we
want
to
do
is
go
through
the
housekeeping
which
there
really
isn't
any
and
just
for
those,
maybe
new
to
the
call
or
don't
and
don't
know
what
we
do.
A
But
essentially
we
have
a
series
of
project
boards
available
under
the
kubernetes
SIG's
repo
or
kubernetes
SIG's
org
for
architecture
tracking.
It's
the
name
of
the
board's
repos.
So
you
can
go
there
and
look
at
sort
of
the
in-flight
things
that
we
have
going.
We
track
conformance
tests,
caps
that
are
in
flight
and
API
reviews
and
we
try
and
keep
those
updated,
and
there
is
a
mention
in
the
agenda
basically
that
this
job
of
maintaining
these
boards
is
getting
a
bit
onerous
for
one
person,
namely
myself.
A
So
we're
going
to
be
looking
at
looking
at
automation
for
those.
So,
if
anybody's
interested
in
talking
to
me
about
helping
with
the
automation
project
there,
I
would
definitely
welcome
that
so
movie,
right
along
cloud
provider,
repos,
there's
a
discussion
about
this
going
in
the
forums
Caleb
miles.
Are
you
on
this
call
I'm
scrolling
through
list
to
see
if
Caleb's
on
and
it's
not
looking
like
it
Brian?
Do
you
have
any
updates
on
this
I
wasn't
sure
what
was
that's.
B
B
The
original
seven
cloud
providers
ended
up
intrigued
because
that's
all
we
had
at
the
time
and
then
we
shut
the
door
on
new
cloud
providers
and
invented
a
mechanism
for
external
cloud
providers
to
be
out
of
tree.
All
of
the
original
seven
are
still
in
the
kubernetes
kubernetes
codebase,
and
there
are
efforts
underway
across
all
of
them
to
extract
those,
and
there
are
benefits
to
doing
this
waiting
for
open
source
kubernetes
to
cut
a
new
released
change
or
fix
something
that
is
cloud
specific.
B
For
example,
the
discussion
that
we
had
in
the
club
provider
sig
was
around
what
we
we
don't
want.
Every
two-person
company
to
be
providing
cloud
providers
in
in
kubernetes
org,
and
so
we
discussed
some
criteria
about
the
amount
of
support
that
it
would
need
to
demonstrate
and
documentation
and
discussions.
B
That
brings
us
up
to
current
state,
at
which
point
now
we
have
kubernetes
SIG's
and
some
of
these
kinds
of
projects
end
up
in
kubernetes,
States
I
think
it's
reasonable
to
have
that
as
a
second
step
after
this
space,
but
I
think
we're
looking
at
I,
don't
know
a
year,
probably
before
all
current
entry
cloud
providers
are
able
to
do
that,
migration,
and
so
looking
at
another
year
where
there
are
there
is
the
potential
for
a
perceived
distinction
between
classes
of
cloud
providers.
So
I
have
a
question
yeah.
A
B
Are
not
among
those
in
it
original
seven?
That
is
the
proposal
is
that
to
create
parity
between
the
entry
cloud
providers
which
moved
from
communities
kubernetes
to
kubernetes
cloud
provider
XYZ
and
create
corresponding
repositories
or
external
cloud
providers
to
be
included
in
that
group
equally
and
is.
A
B
A
B
One
requirement
or
an
aspirational
goal
or
in
the
future,
but
we
did.
We
did
discuss
that
as
requirement.
I
think
OpenStack
is
doing
it
now
and
digitalocean
is
working
on
it.
Maybe
that
again
there
are
benefits
to
doing
that.
So
we
don't
think
we
need
the
stick
on
that
to
become
aware
that
some
change
breaks
for
cloud
provider
is
motivational
incentive.
B
B
B
A
B
A
D
On
this
is
Tim
on
the
mailing
list.
I
was
arguing
that
these
repos
should
not
go
in
companies
that
she
won't
come
86,
not
realizing
that,
actually
that
we
had
already
started
the
process
movie,
you
know,
and
even
though
those
the
existing
ones
are
covered
under
the
grandfather
clause.
I
don't
feel
like
it's
fair
to
new
cloud
providers.
D
B
I
think
it's
around
an
escalope
response,
time
and
timeliness
about
applying
certain
patches
I.
We
will
because
an
action
item
to
send
that
to
the
thick
art,
mailing
list
and
I'm,
also
okay,
with
creating
right
now
and
a
goal
as
a
set
as
a
subsequent
phase
to
move
them
all
out.
At
the
same
time,
all
right
I
think
it's
reasonable
to
take
on
that
extra
work
in
the
future.
If
it's
useful
to
me
get
through
this
space.
E
B
D
The
same
thing
I
mean
Mike,
you
can
talk
last
year,
basically
proposed
kicking
everybody
out
of
the
tree
and
even
out
of
the
release
time
right.
If
we
that's
a
pretty
major
decision
that
changes
for
the
entire
operating
mode
of
what
we
release
every
quarter,
I
don't
think
we're
ready
to
go
there.
Yet
there's
an
issue
also
around
legal
concerns
out
repatriating
typewriter
code
into
nonterminating
depositories,
where
like,
for
example,
the
google
CLA
is
not
the
same
as
the
cnc
MCLA,
even
though
they
are
very
similar.
A
Licensing
here,
because
we're
not
lawyers
there
and
and
believes
that
what
convention
is
really
in
issue
we'll
have
to
discuss
that
what
people
are
actually
I
can
speak
to
that
we
have
so
DIMMs
had
his
hand
up
and
Brad
had
they
set
up
as
well.
So
let's
maybe
give
them
the
floor
for
a
second
Tim's.
Okay,
thank.
C
C
If
we
bring
the
new
ones
like
digital
provider
into
digital
ocean
into
Cuban
8:06,
then
we
are
setting
up
two
different
classes
there,
because
there
are
some
cloud
providers
in
curator's
org,
some
in
Cuban,
it
is
say
so
I
think
to
be
fair
to
everyone.
What
we
should
be
doing
is
move
all
the
cloud
providers
into
Cuba
notice,
FIGS
eventually.
C
D
Think
that
is
exactly
what
we
are
deciding
right:
okay,
we
can
rule
out
two
ocean
etc
into
kubernetes.
Today,
based
on
the
rubric
to
be
decided
by
sig
provider
and
a
possible
eventual
state
would
be
to
move
them
all
to
a
different
organization,
whether
that's
communities
saves
or
communities,
cloud
providers
or
something
else,
yeah.
C
B
Back
to
the
cloud
provider
now,
I
think
not
everyone
who
has
an
opinion
on
this
is
here,
but
we
understand
the
goal
to
be
that
we
all
end
up
in
community
six.
We
also
understand
there's
additional
work
to
move
twice
and
potentially
issues
with
discoverability
and
dead
links
and
all
kinds
of
things.
So
we
will
take
this
back
and
understanding
that
that
is
the
eventual
goal.
We
will
debate
the
best
way
to
get
there
with
the
agreement
of
this
group
that
we
can
have
an
incremental
step.
If
that
is
necessary,
is
that
good,
summary
discoverability.
D
C
D
A
E
Yeah,
just
an
update
on
this
topic
from
the
from
the
sig
dock
dock
maintainer
view
of
the
same
issue.
You
know
one
of
the
things
one
of
the
problems
we
had
and
I'm
people
know
we're
doing.
A
restructuring
of
the
dock
content
and
you've
got
the
in
cloud.
You've
got
the
entry
providers
or
what
have
you
and
the
navigation
is
a
mess
and,
at
the
same
time
we
had
several
folks
wanting
to
update
their
offerings.
E
A
lot
of
people
change
their
names
from
an
idea
from
a
from
a
container
service
offering
to
a
kubernetes
service,
offering
I
think
there
were
at
least
three
different
folks
that
have
changed
their
names
to
highlight
the
fact
that
their
Nettie's,
which
is
really
nice.
So
what
we
did
is
a
compromise
because
you
know
we
had
all
these
issues
and
we're
coming
back
with
an
answer
that
says:
well,
we
were
structuring
the
docs,
so
hold
off
hold
off,
but
at
the
same
time
many
of
the
offerings
were
like
I,
really
want
to
update.
E
The
information
about
my
my
offering
and
my
kubernetes
offering
we
resolved
to
do
was
used
to
picking
the
right
solution
page
as
sort
of
at
least
the
one
page
that
we
were
trying
to
keep
up-to-date
with
the
updated
information
about
all
the
different
kubernetes
kubernetes
offerings
from
the
vendors.
So
at
least
there
was
one
page
that
was
like
a
record
of
truth,
while
all
this
restructuring
goes
on
in
both
the
doc
navigation.
And
what
have
you
so
I
just
want
to?
C
A
Okay,
moving
along
in
the
agenda,
just
a
quick,
please
take
a
look
note
in
the
agenda
about
stephen
augustus,
has
a
update
to
the
caps
process
and
just
everybody.
If
you
could
just
take
a
look
at
that
and
give
you
two
cents
that'll
be
great.
Yeah
just
could
be
out
on
that.
It's
basically
this
stuff.
We
discussed
previously
this
meeting
and
an
email,
threads
and
issues
about
moving
the
caps
and
reorganizing
kept
some
design
proposals
and
unifying
caps
and
feature
process
like
that.
It's.
A
A
And
we
had
a
discussion
about
feature
gate
and
the
deprecation
policy
for
alpha
feature
gates.
There's
document
here,
that's
been
in
progress
and
needs
review
I.
Don't
necessarily
think
we
need
to
review
this
in
this
meeting,
but
I
believe
it
is
something
that
needs
some
eyes
on
it.
Yeah.
Actually
one
thing
I
would
like
to
do.
A
You
know
we
haven't
really
got
the
cap
or
API
review
processes
moving
and
the
conformance
reviews
so
far
just
been
a
couple
of
people
doing
them
well,
I
think
one
thing
that
may
help
is
actually
assigning
owners
to
some
of
these
things.
They're
like
make
somebody
responsible
for
pushing
it
over
the
finish
line.
So
does
anybody
feel
passionately
about
the
obligate
issue
that
would
be
willing
to
volunteer
to
pick
it
up
and
sort
of
drive
it
to
closure?
I.
A
Think
the
people
who
might
be
passionate
about
it
are
not
your
things,
I
think!
That's
right!
Actually,
so
we
can
put
that
on
the
mailing
list
or
we
can
actually
I'll
do
it
in
slack
and
we'll
just
put
it
up.
There's
somebody
needs
to
claim
this
or
it's
simply
not
gonna
move
forward.
Could
you
do
it
by
a
mailing
list
like
I?
Just
can't
I
pull
slack
roughly
once
a
week.
So,
okay,
you
know,
let's
do
this
thanks.
A
Lastly,
there
is
a
comment
regarding
the
the
profiles
for
conformance
Brad,
if
you're
still
on
the
call
he
had
raised,
but
concern
basically
that
there's
some
consternation
between
the
CNCs
view
of
the
testing
and
what
we're
trying
to
do
with
profiles
and
I.
Think
Brad
just
wanted
some
clarification
on.
Why
we're
looking
at
profiles
and
what
the
purpose
is
and
what
ways
excused
from
the
notion
of
conformance
as
the
CNC
up
sees
it
yeah,
both
Aaron
Berger
and
Timothy
Sinclair
convinced
in
today,
so
I
pushed
it
to
next
week.
A
A
Just
a
quick
public
service
announcement
be
a
good
citizen
and
involve
yourself
in
the
election
process
for
the
steering
committee.
So
that's
coming
up
next
week
will
be
the
votes
the
will
be
sent
out
for
the
service
voting
process.
So
please
to
vote.
We
have
some
really
good
candidates.
Dimms
is
one
of
those
Greek
units,
so
let's
definitely
get
out
there
and
get
our
next
wave
of
leaders
installed.
A
C
A
Proactively
earlier
planning
ahead,
for
things
start
getting
resolved
and
email
or
you
know,
just
if
you
want
to
get
more
input
for
more
people
than
usual
suspects,
definitely
feel
free
to
try
to
get
things
on
the
agenda.
I
think
the
dock
is
writable,
or
these
comments
look
like
furniture,
so
anybody
can
edit
it.
So
the
the
corollary
to
that
too
is-
and
this
is
something
I
think
we
need
to
be
a
bit
more
strict
on-
is
if
there
are
no
agenda
items
say
a
day
before
the
meeting.
A
We're
gonna
cancel
the
meeting,
because
I
don't
like
having
things,
get
dropped,
dropped
in
the
last
minute
and
people
don't
have
time
to
review
them.
So
it's
really
in
our
fishing.
So
like
there's
a
24
hour
window
there,
where,
if
there's
nothing
in
the
agenda,
then
there's
a
very
high
likelihood
of
cancelling
yeah
yeah,
plus
one
of
that
for.