►
From YouTube: kubeadm office hours 2020-09-16
A
A
A
Okay,
so
I
don't
have
any
psas
for
today.
I
think
something
important
to
notice
here
is
that
we
are
deprecating
docker
shim
in
kubernetes.
B
A
See
basically,
the
pr
wanted
to
deprecate
docker
shim
without
a
cap,
but
we
decided
that
the
cab
is
definitely
needed
here.
B
A
I
don't
think
we
should
focus
too
much
on
it
because
we
don't
have
that
much
time,
but
I
just
wanted
to
focus
on
some
of
the
responses
and
I
extracted
some.
You
know
trying
to
collect
some
themes
for
from
the
survey
about
kubernetes.
A
Okay,
so
from
72
users
we
got
only
40
percent
that
want
to
provide
feedback.
That's
29
people,
the
overall
ux
cubed
m
experience
is
eight
out
of
10
average,
which
is
good
a
lot
of
users
from
production
coasters
with
kubernetes.
A
A
Hs
support
doesn't
seem
to
be
that
important
for
some
people
overall,
this
is
quite
sparse.
Upgrades
seem
to
be
quite
important.
A
This
also
is
not
that
deterministic
for
a
customization,
static
ports,
image
prepo,
face
support,
are
also
quite
the
non-deterministic
force.
Resetting
the
load
seems
to
be
important
for
of
people
and
also
not
so
important
for
all
people
rate
the
following
areas
of
kubernetes
based
on
how
easy
they
were
for
you.
A
So
it's
not
like
kubernetes
easy
for
a
lot
of
people
using
the
cli,
also
using
the
config
setup
in
the
couplet
dns
proxy
container
on
time
those
seem
to
we.
We
basically
want.
We
want
users
to
to
have
a
five
here.
A
Basically,
we
want
this
to
be
super
easy
for
the
users,
but
some
of
these
are
also
not
really
the
responsibility
of
kubernetes
in
you
know,
if
you
look
at
cri,
for
instance,
kubernetes
doesn't
really
set
up
the
cri
for
you.
A
And
these
really
don't
give
us
that
much
detail
managing
certificates.
Upgrades
asia
supports
its
investment
again.
We
want
here
to
to
make
this
as
easy
as
possible,
so
potentially
more
people
voting
for
a
five
here.
A
I'm
surprised
that
we
have
a
lot
of
users
that
do
that,
I'm
going
to
assume
that
I
know
wildly
assumed
that
these
are
not
quasar
api
users,
so
maybe
we
have
users
that
don't
basically
that
they
treat
most
as
immutable.
An
action
item
here
is
that
we
should
potentially
attend
to
and
test
for
immutable
node
upgrades.
B
Within
are
you
basically
saying
throwing
away
the
you
know,
then
creating
a
new
one.
B
A
That
is
true,
I
guess
so
we
can
based
on
cluster
api
tests.
We
can
stop
doing
that.
I
personally
do
not
know
where
the
links
are
for
that
like
where,
where
is
the
test
grid?
For
that?
Is
it
green?
I
don't
know.
B
Oh,
it
is
in
in
the
c
cluster
life
cycle,
there
is
a
cluster
api
yeah.
To
be
honest,
there
is
a
cluster
api.
B
B
Also,
these
are
upgrade
tests
yeah
these
basically
in
copy.
We
have
only
one
job
that
that
groups
different
end-to-end
tests
and
there
is
kcp,
upgrade
machine
deployment
upgrade,
and
so
there
are
three
upgrade
testing
in
this
street.
B
B
B
A
Yeah,
okay,.
A
I
mean
we
could
discuss
this
more
eventually.
I
don't
know
it
might
be
a
good
idea
to
have
separatists.
A
We
might
be
able
to
help
other
six
by
testing
component
skew
and
things
like
that
as
well.
Extending
kinder
so.
A
Pending
discussion,
I
guess
kubernetes
operator
a
lot
of
people
said
they
are
interested
about
it.
Some
people
say
that
they
are
not
sure
this.
This
is
a
question
about
the
order
of
importance
of
features
that
the
kubernetes
operator
is
going
to
have
higher.
So
is
one,
so
I
guess
provide
high
level
abstraction.
How
hide
away
details
is
something
that
people
want
also
observability.
A
It
seems
that
people
are
not
that
interested
in
extensibility
here,
which
is
yeah.
A
Yeah,
I
guess
so
what
type
of
node
changes
you
want
the
operator
to
perform.
The
weather
here
is
reconfigure.
The
control
plane,
reconfigure
nodes
is
not
that
preferred
here
and
ca
rotation,
certificate,
rotation
and
core
add-on
configuration
is
least.
A
Preferred
so
potentially
we
can
call
the
operator
do
this
as
the
like.
The
primary.
B
A
A
B
From
mesosphere,
so
they
are
the
daniel
and
they
ask
them
many
times
in
for
mutable
upgrades
or
multiple
changes
during
the
sick
cluster
api
meetings.
A
Yeah,
I
see
it's
definitely
also
a
nice
action
to
for
the
operator
to
support.
I
guess
a
possibly
higher
higher
requested
action
compared
to
reconfiguring
the
control
plane
like
it's
probably
the
most
requested.
A
A
So
these
are
direct
responses
about
what
is
the
most
desired
feature.
I
think
I
gathered
some
a
summary
here,
so
this
is
about
operator.
I
think
there
are
hints
about
the
documentation
of
the
around
the
kubernetes
config
and
I
have
a
issue
walk
to
the
kubernetes
website
to
improve
our
links
a
little.
How
do
we
link
to
the
the
cube
adm
api
page?
I
think
people
still
confuse
where
to
find
it.
B
Yeah,
I
kind
of
think
that
we
have
to
bring
back
the
api
documentation,
if
not
the
api
documentation,
at
least
one
example,
one
or
two
examples.
So
what?
What
is
the
the
docs
dot
go
from
the
api,
which
is
just
an
introduction
and
then
link
to
the
api
for
the
object?
So
the
user
at
least
gets
few
working
examples
and
few
pointers
on
how
to
go
into
the
godoc.
A
Yeah
I
I
agree.
This
has
been
requested
like
every
release,
pretty
much.
We
we
don't
have
user
stories,
we
don't
have
examples.
Maybe
we
should
just
open
the
document,
create
a
couple
of
stories
and
wait
for
people
to
start
contributing
there.
B
A
Yeah
we
could
just
just
retitle
the
api
version
and
potentially
remove
examples
that
are
no
longer
valid
for
this
api
and
yeah.
That's
it.
I.
I
think
that
I
should
take
an
action
item
for
that
and
let
me
try
to
assign
myself.
A
Yeah,
so
a
lot
of
areas
incubation
such
as
can
still
reportedly
be
not
so
easy
a
few
comments
about.
So
I
saw
this
in
the
replies
here
that
people
for
some
reason
want
cube
adm
to
concurrently
create
the
control
plane,
not
by
basically
people
questioning
the
way
cuba
dm
is
designed
with
the
init
slash
join
workflow.
A
They
want
the
control
plane
to
be
created
at
once,
so
there's
no
leading
primary
control
plane.
I
I'm
pretty
sure
cops
works
that
way
you
just
request
the
machines,
but
this
is
a
very
complicated
change
and
you
know
it's
a
fundamental
change
in
the
way
kubernetes
designed.
C
A
C
A
I
I
think,
but
that's
like
the
primary
reason
I
think
cops
does
not
still
want
to
move
to
cuba.
Dm
is
because
of,
I
don't
think
they
want
to
handle
the
race.
But
you
know
this
is
a
different
discussion.
D
A
Guess
yeah,
so
please
have
a
look
at
all
the
comments.
Some
of
those
are
not
really
that
relevant,
for
instance,
their
requests
for
features
that
are
completely
out
of
scope
for
kubernetes.
A
I
believe
there's
one
comment
that
kuberium
is
too
complex
and
it
just
wraps
shell
scripts,
which
is
not
exactly
true
comment
about
h,
a
b
being
too
complex.
A
A
Yeah,
so
this
is
pretty
much
the
summary.
I
encourage
you
to
have
a
look
at
the
whole
survey.
There
are
interesting
points
about
the
state
of
upgrades
like
how
how
much
behind
people
are
following
the
kubernetes
recycle,
you
have
cost
api
feedback
and
also
lcd
feedback.
A
Any
more
comments
for
the
survey
we
should
move
to
the
planning.
We
have
40
minutes.
A
A
We
have
the
120
one.
Basically
next
milestone
that
I
extracted
ideas
from
the
important
dates
you
can
for
the
120
release
are
here.
A
So
the
first
important
date
for
us
is
in
husband
phrase
which
is
october,
the
6th,
which
basically
means
we
have
something
like
a
couple
of
months.
A
Actually,
one
month,
sorry
before
we
write
any
caps
and
merch
so
yeah.
This
is
a
the
capstone
one.
B
But
and
maybe
I'm
wrong,
but
I
don't
see
new
caps,
what
couldn't
mean
going
on
the
cycles.
A
Potentially
we
can
edit
about
the
customize
deprecation
we
can
edit
the
patch
escape
and
remove
the
old
cap,
or
at
least
remove
the
tracking
issue.
We
shouldn't
remove
the
cap.
We
can
remove
some
tracking
issues
and
make
adjustments,
but
that
is
it.
I
don't
think
we
have
new.
B
Cabs,
it
would
be
nice
to
to
update
the
club
to
come
to
describe
the
current
solution.
A
Okay,
so
this
is
an
item,
but
it
potentially
it
contains
a
lot
of
different
items.
It's
about
deprecations
removals.
A
Yesterday
we
had
a
discussion
about
the
first
one,
which
is
removing
v1
beta1
question
api
prefers
that
we
don't
remove
it
yet
because
there's
some
lack
of
clarity
around
how
to
manage
multiple
versions
of
the
api
yeah.
There
are
definitely.
B
A
Yeah,
do
you
think
that
this
these
dates
are
going
to
synchronize
if,
for
instance,
if
21
is
released
in
march,
do
you
think
that
you're
going
to
have
a
version
that
supports.
B
A
Yeah,
and
also
something
to
measure
here
is
that
I
believe
team
hawking
proposed
that
beta
apis
should
be
deprecated
after
four
releases,
instead
of
three
as
the
minimum.
A
So
we
are
going
to
follow
the
for
release,
duplication
policy
if
we
depict
if
we
remove
in
121.
A
Okay
moving
to
the
next
one,
so
here
I'm
going
to
say
pointing
to
21.
A
This
is
up
something
that
is
scheduled
for
a
removal
in
120.
This
is
a
flag
that
we
have
it's
a
pretty
easy
change.
I
think
it's
just
it's
scheduled
for
120,
it's
a
low
priority,
I
think
mostly,
so
maybe
we
can
have
somebody
help
with
that.
That's
one
for
me
for
the
package.
A
This
is
deprecation.
Removal
of
the
experimental,
dynamic,
complete
configuration
support.
I
I
set
the
pr
last
cycle
to
deprecate
it
that
is
scheduled
for
removal
in
120.
The
pr
is
already
out,
I
think,
for
virtually
lgbt.
I
think
already.
A
Replace
customize
withdrawal
patches-
this
is
scheduled
for
120,
so
do,
for
which
we
have
any
objections
around
that
we
had
one
release
of
deprecation.
B
I'm
okay!
I,
if
I'm
not
wrong
there
was
someone
in
the
cluster
api
community
using
that
maybe
people
from
metal
cube
prakesh,
if
I
remember
where,
but
but
I
I'm
not
sure
so-
let's
bring
in
the
cluster
api
channel
just
for
security,
but
I'm
plus
one
for
the
application.
A
Yeah
there's
there
are
also
people
with
vmware
that
are
using
customize
in
one
of
the
products
experimental
stuff.
Basically
I
told
them
that
they
should
move
to
patches
and
they
say:
okay,
it's
fine.
We
can
branch
between
the
releases,
the
other
release.
Can
you
customize?
The
release
can
use
patches.
They
said,
there's
not
a
problem
for
them.
B
Okay
right,
so,
let's
just
inform
the
in
the
in
the
channel
that
or
let's
do
this
way
now
we
have
a
sitting
what
we
are
going
to
duplicate
and
then,
let's
communicate
at
least
of
everything
we
are
duplicating
as
broad
is
to
broadcast
this
information
as
much
as
possible.
A
So
this
is
something
that
is
left
over
from
us
last
release,
yet
richard
said
some
pr's,
but
we
decided
to
not
deprecate
this
particular
flag.
I'm
not
sure
this.
Actually,
this
is
scheduled
for
122,
because
it's
ga,
so
I
I
I
guess
I
have
to
change
the
milestone
here.
Okay,
next
yeah,
I
I'm
going
to
punch
this
one
because
it's
not
clear
yet.
A
This
is
also
not
clear.
We
we
don't
know
if
we're
going
to
have
to
remove
the
the
insecure
port
for
the
api
server.
I'm
going
to
leave
this
without
a
decision
here,
because
we
just
don't
know
we
have
no
idea.
If
this
is
going
to
happen,
I
I
I
propose
that
we
deprecate
the
self-hosting
pivot
functionality
in
kubernetes.
A
A
A
A
potential
action
item
for
me
here
is
to
ask
you
know:
arafa
and
marek.
A
But
I'm
plus
one
to
deprecate
self-hosting
support.
C
Yeah
a
long
time
ago,
we
asked
for
users
to
join
this
call
and
advocate
for
this
to
remain
in
kubernium.
We
kept
it
because
some
people
did,
but
it's
been
a
long
time
since
then
and
usage
has
not
increased
and
there
are
other
projects
that
do
this
kind
of
thing.
C
I
know
that
talos
systems
moved
to
boot,
cube
instead
of
using
kubernetes
self-hosted
features,
so
this
makes
sense
for
a
deprecation
plus
one.
A
A
So
that's
about
the
applications.
I
mean
I'm
going
to
probably
focus
on
this
this
this
cycle,
I'm
going
to
work
on
deprecations.
This
is
my
goal.
We
have
too
many
features
that
are
stale
or
we
have
to
move
the
patches
forward.
We
have
to
duplicate
the
couplet
dynamic
config.
A
This
is
like
I'm
planning
this
to
be
my
primary
assignment.
I
may
ask
for
help
somewhere.
A
You
know
just
look
at
the
tickets
and
offer
help
if,
if
you
want
overall,
the
this
is
something
like
a
p3
or
p2,
I
guess
deprecations.
A
B
I
agree:
we
had
a
good
feedback
from
the
from
the
users,
and
so
I'm
plus
one.
A
At
this
point,
the
alpha
has
no
significance,
because
the
feature
has
been
out
there
for
a
while.
We
don't
plan
to
make
any
breaking
changes
to
the
commands.
I
think
so.
I
think
it's
fine
to
just
move
it
up,
but
my
proposal
here
is
to
have
the
two
commands
and
in
parallel,
what
do
people
think
so
we
can
have
kubernetes
alpha
cert
and
also
we
can
copy
the
command
to
kubernetes
search
for
one
release.
C
We
should
just
post
a
deprecation
notice
in
the
release
and
then
also
in
the
command
output
or
something
your
standard.
C
A
Okay,
does
anybody
want
to
take
this
action.
A
Okay,
thanks,
let
me
actually
let
me
give
this
something
like
a
so
I
guess
this.
This
should
be
something
like
important
zoom
because
of
the
release,
maybe
important,
lockdown,
because
it's
deprecation
in
the
next
release.
A
Let
me
remove
myself
from
the
top
from
here:
okay
graduate
kubernetes
buffer
config
user.
So
this
is.
This
is
a
bit
problematic,
because
the
way
the
way
this
command
works
is
that
it
can
be
used
to
generate
cube,
config
files
that
are
compatible
with
cube
adm.
A
A
B
B
So
this
is
something
separated.
I
don't
have
a
strong
opinion
on
on
graduating
or
not.
E
I'd
say
that
we
should
probably
postpone
it
a
little
bit
and
see
the
exact
research
like,
for
instance,
just
ask
some
users
in
the
main
place
to
somewhere
to
see
exactly
what
the
ux
should
be
here.
A
Yeah
this
this
about
has
been
alpha
for
a
very
long
time.
I
think
it
has
been
since
maybe
1
11
or
something
like
that.
A
It
has
been
used
by
a
lot
of
people
and
there
are
a
lot
of
users
requesting
that
it
shouldn't
be
alpha,
because
some
people
don't
want
to
use
alpha
commands
when
they
see
alpha,
they
don't
see,
they
don't
want
to
use
it
because
it
can
break
that's
their
concern.
D
A
Yeah,
I
I
think
one
problem
with
the
existing
combined
is
that
it
it
accepts
a
bunch
of
flags,
and
maybe
we
want
to
make
it
accept
a
coaster,
configuration
and
flags
for
only
those
fields
that
are
not
present
in
the
quasi-configuration,
like
a
common
name,
for
instance,.
A
A
A
Okay,
I
also
added
a
topic
about
v1
beta
3.
So
what
what
concern
that
I
have
with?
Basically
just
looking
at
the
issue
track
that
we
have
in
cuba
in
the
cubed.
C
A
Is
that
the
list
is
growing
of
api
changes
and
at
some
point,
there's
going
to
be
so
much
volume
to
you
know
all
these
small
changes
that
we
have
to
add
to
the
new
api
is
that
we
are
not
going
to
have
the
bandwidth
to
execute
this
in
the
same
cycle,
especially
given.
Now
that
a
lot
of
folks
don't
actively
contribute
to
kubernetes.
A
This
is
my
concern.
I
I
don't
think
we
can
wait
much
longer
for
or
we
have
to
selectively
pick
features
that
are
going
to
land
in
v1
beta3
and
postpone
some
features
to
the
next
api.
A
So
again,
there's
a
problem
of
bandwidth,
so
I
don't
think
we
can
execute
if
you
beta
3
in
this
cycle.
But
what
what
do
people
think
like
when,
when
should
we
say
like
okay,
like
122,
is
the
the
release
where
we
are
going
to
bump
the.
E
Api,
I'm
thinking
here
that
we're
actually
using
the
api
change
label
too
often,
and
if
we
actually
go
into
every
single
one
of
those
issues,
we
might
actually
get
this
list
to
be
at
least
half
the
size.
That
is
currently
so
probably
just
a
few
things
here
worth,
including
in
a
new
beta
api
version.
For
example,
the
object
method
of
the
configurable
sandbox
issue
like
sandbox
post
image,
is
probably
not
going
to
end
in
there.
At
least
I
am
against
it.
E
So
there
are
a
lot
of
stuff,
that's
that
are
questionable
and
probably
should
not
land
in
a
new
version
here,
but
if
we
actually
use
sort
of
like
a
web
pr
that
acts
like
a
say,
a
branch
in
which
to
develop
better
three.
We
can
probably
just
hold
it
off
for
this
cycle
and
merge
it
early
in
the
next
one,
but
still
use
this
cycle
as
a
development
cycle.
B
A
B
I
I
will
be
fine
to
have
to
agree
on
rpi
design
and
and
then
be
ready
to
start
at
the
beginning
on
the
next
cycle.
A
Also
question
here
is:
do
do
we
really
want
to
change
to
have
major
changes
in
the
api
or
just,
for
instance,
v1
beta
3
can
just
be
a
cleanup.
For
instance,
we
have
to
remove
cost
steros.
A
A
new
feature
is
skipping
phases.
The
question
api
have
been
requesting
this
for
a
while
remove
dns.
These
are
minor
things
and
the
question
here
is:
should
we
not
even
design
the
new?
I
just
add
these
and
remove
these
minor
things
from.
A
A
Yeah,
I
guess
also
we
should
probably
start
with
the
design
which,
which
can
be
a
change
to
the
cap,
enumerating
what
what
we
want
to
remove.
What
we
want
to
add.
I
general
I,
I
I
don't
think
what
people
think
about
the
idea
of
incrementing
the
api.
More
more
often
before
it
graduates
to
ga,
we
can
have
many
versions
of.
B
A
discussion
on
what
will
be
the
future
occurred
mean,
and
but
I
I
will
take
it
and
this
and
this
discussion
regards
also
the
the
api,
but
I
I
will
trigger
this
discussion
after
we
go
through
all
the
all
the
items.
If
there
is
time.
A
Yes,
something
that
I
I
liked
about
the
office
is
that
we
cremated
much
faster.
We
we
had,
I
think
one
of
the
releases
alpha
2
was
a
pretty
pretty
minor
change
compared
to
v1
alpha
1..
A
I
don't
think
we
should.
I
think
we
should
move
faster
with
the
api.
Basically
add
new
versions
remove
old
versions,
even
if
we
don't
have
enough
volume,
because
I
don't
think
it's
going
to
be
very
difficult
for
whoever
is
he
is
babysitting.
That
big
pr
is
is
it
might
be
a
bit.
A
Difficult
okay!
So,
for
this
cycle,
do
we
does
everybody
want
to
enumerate
the
the
changes
of
potentially
the
next
iteration
of
the
api?
Or
just
do
we
want
to
just
hold.
B
A
Oh
yeah
yeah,
basically
the
the
labels
are,
is
accurate.
The
api
change
our
label
is
accurate
everywhere.
Here
I
think
what
we
should
do
is
basically
take
this
v1
beta3
config
enumerate
all
the
changes
that
we
want
to
make.
This
is
something
that
we
can
do
in
the
next
meeting.
I
guess.
A
A
C
Hi
friends,
rajan
sandeep
from
the
cluster
add-ons
call
brought
up
yesterday
that
he
was
curious
about
talking
just
just
making
the
add-on
situation
a
little
bit
moving
that
forward
or
sometimes
a
cycle,
but
he
wasn't
able
to
make
it
today
and
then
I
believe
some
touchy
from
the
cluster
islands
group
is
also
on
with
us
and
yeah.
I'm
open
to
have
conversations
about
this
or
set
up
a
design
meeting
later
on
at
a
different
time.
A
A
I
I'm
monitoring
the
situation
in
cops.
They
are
implementing
the
operator,
I'm
also
watching
the
tracking
issue
of
the
addons
repository
for
the
coordinates
operator.
A
I
I
would
like
personally
to
see
how
how
things
go
in
cops-
and
maybe
this
was
my
proposal
last
time-
cube
adm
can
basically
add
the
operator
instead
of
adding
the
deployment
and
managing
the
deployment.
This
was
my
proposal
from
last
time.
I
quite
frankly,
I
I
don't
even
know
why
we
blocked
on
that.
I
know
that
some
of
the
the
picture
around
upgrades
is
not
very
clear
around
the
crd,
but.
A
C
That
didn't
get
done.
Okay,
I
know
that
that
was
a
conflicting
area
of
work
from
the
add-ons
patch.
Okay
do
as
a
separate
kind
of
feature
consideration
because
yeah
there's
there's
the
one
item
about
moving
coordinates
operator
forward.
Do
we
want
to
add
the
ability
to
disable
kubernetes
usage
of
add-ons?
I
know
that
this
is
possible
in
a
bespoke
manner
by
using
phases,
but.
A
I
I
think
it's
possible
today,
for
are
you
talking
about
upgrades,
not
applying
coordination
upgrade.
A
A
So
I
think
cuban
m
today
already
supports
the
lack
of
coordinates
from
the
cluster.
A
Yes,
you
can
also
skip
q
proxy
the
same
way
rusty.
Can
you
confirm
my
comment.
E
Yeah,
as
far
as
I
remember,
that
was
the
case,
and
also
there
is
at
least
I
think
that
there
is
an
issue
about
adding
sort
of
a
field
to
the
config
and
battery
that
is
going
to
allow
us
to
list
what
add-ons
to
be
installed.
E
That
probably
should
get
checked
and
see
whether
it's
going
to
one
is
better
three
and
like
stuff
like
that.
So
the
idea
there
was
to
have
a
little
bit
nicer,
ux
and
just
go
and
skip
a
face
somewhere,
but
the
idea
initially
was
to
just
keep
out
of
the
way
of
the
coaster
addons
and
just
allow
them
to
be
deployed
manually
by
the
users
and
in
the
future,
even
by
cubed
situation.
C
Yeah,
this
ability
for
for
people
to
skip
phases
when
we
don't
provide
fields
is
very
useful
and
they
think
increasingly
important
in
an
environment
where
we
have
people
who
are
maintaining
replacements
for
components
like
cooproxy
as
well
as
more
interesting
multi-cluster
topologies
that
can
rely
on
external
coordinates
the
I
guess
the
only
other
question
I
have
on
this,
because
I'm
pretty
satisfied,
you
know,
since
we
have
the
ability
to
have
the
user
opt
out
of
cuba.
Damn
managing
add-ons
is.
C
Do
we
have
the
api
field
inside
of
our
component
configs
for
the
kubernetes
cli
that
allow
people
to
skip
phases
via
the
api
objects.
A
A
Oh
for
british,
I
just
I
if
people
decide
for
this,
I'm
fine,
but
I'm
personally,
minus
one,
because
it's
a
major
pain
to
try
to
coordinate
with
users
in
what
version
a
certain
field
was
at
it.
It's
just.
I
don't
want
to
deal
with
that
on
the
communication
side
between
developer
and
user,
just
all
right.
It's.
A
Okay,
in
general,
other
areas
of
kubernetes
are
perfectly
fine
with
this,
but
you
just
have
to
search
when
this
api
field
was
added
and
it's
as
somebody
who
talks
with
users
every
day.
I
I
don't
want
to
get
into
this
business.
Somebody
else
raise
their
head.
E
D
E
C
Oh
okay,
yeah
the
as
far
as
the
adding
the
api
field
thing.
If
that's
the
only
reason
you
know,
we
could
just
add
a
comment
in
the
api
on
the
field
that
documents
when
it
was
added.
A
Yeah,
we
should
definitely
iterate,
like
I
said
earlier,
I'm
perfectly
fine
to
iterate
the
view
on
a
beta3
just
to
support
one
feature.
I
don't
think
we
should
collect
volume
to
release
apis,
but
you
know
if
this
blocks
a
lot
of
people
like
this
feature,
I'm
perfectly
fine
to
just
release
a
small
api
increment,
maybe
zoom
is
booked.
I
see
people
raising
heads.
B
A
Well,
basically,
territory
of
the
cap
with
the
idea
to
support
a
list
of
adults
inside
the
api.
C
B
The
option
of
disablement
is
just
intended
to
allow
experimental
around
their
dawn's
operator
idea,
okay,
but
but
once
the
the
experimenting
is
finished,
we
should
make
this
a
more
robust
solution
so
that
the
user
can
consume
in
an
easier
way
in
kubernetes
and
and
these
require
the
cap
to
get
finished
so
to
address.
I
don't
know
how
to
get
the
list
of
images,
how
to
manual
upgrade
how
to
how
to
allow
users
to
customize
things
and
so
on
and
so
forth.
B
So
there
are
a
lot
of
tennis
questions
there,
also
in
terms
of
guarantee.
So
we
what
what?
What
is
the
list
of
add-ons,
which
are
we
we
are
starting
to
rely
on
and
so
on
and
so
forth.
So
it
is
just
it's
just
a
two-step
approach.
One
is
we
make
it
possible
to
test
and
to
experiment
around
the
addon
project,
the
second
one
we
we
get
at
the
dome
project
as
a
first
citizen,
the
incubator.
B
C
Sorry,
I'm
just
making
sure
that
that
there
weren't
any
open
questions
on
this
kept.
That
weren't
responded
to.
C
D
B
Yeah,
if
everyone
agree,
I
would
like
just
to
kick
off
a
discussion
that
we
for
the
future
of
kubernetes,
because
immediately
looking
at
the
survey
looking
at
the
feedback
that
that
we
have
at
the
current
discussion
that
we
just
finished
in
my
opinion,
we
should
start
through
kind
of
designing
a
roadmap,
for
the
future
couldn't
mean,
I
think,
will
mean
that
kubernetes,
in
my
opinion,
should
go
through
some
improvements
in
order
to
embrace
the
direction
that
the
community
is
taking.
Barista
and
kubernetes
should
became
more
guitar
friendly.
B
He
could
he
should
embrace
the
operator
for
all
the
dawns
for
for
everything
issue.
It
should
move
to
away
from
the
kubernetes
config
and
and
implement
a
real
crd.
Now
the
crd
are
other
things
in
the
cluster,
so
I
have
in
mind
a
series
of
changes.
I
would
like
to
start
a
discussion
within
this
team
and
then
move
the
discussion
on.
I
don't
know,
write
a
blog
in
order.
B
We
can
advise
the
community
to
to
what
is
that
the
kubernetes
long-term
direction
and
and
start
getting
more
people
to
contribute,
because
there
are
opportunity
and
room
for
making
kubernetes
better.
In
my
opinion,
so
if
you
agree,
I
would
like
to
start
I
don't
know
already
or
to
write
in
a
document
or
to
carve
out
some
time
in
this
meeting
every
every
week
or
every
weeks.
C
C
B
A
Folks,
we
are
out
of
time,
prakasha
check
the
chat.
I
answered
your
response.
You
can
skip
pre-flight
checks.
We
have
to
leave
the
room
to
on
our
coaster
api
to
have
their
meeting
thanks.