►
From YouTube: SIG Cluster Lifecycle - kubeadm office hours 2021-07-07
A
A
We
wanted
to
add
a
couple
of
features
in
this
release,
and
this
was
the
new
api
which
is
v1
beta
3,
and
we
pretty
much
completed
most
of
the
tasks
for
it.
A
One
pending
item
is
the
documentation
update
in
the
kubernetes
website
to
point
users
to
use
the
new
api
as
the
preferred
version.
At
this
point,
as
you
can
see,
we
have
a
lot
of
features
you
can
check
them
out.
Basically,
I'm
saying
this
for
the
recording,
whoever
is
watching
and
for
the
folks
present
on
the
call
we
dropped
a
few
features
and
because
the
picture
was
not
very
clear
around
them,
the
first
one
we
dropped
was
the
cobra
run
directory.
A
A
Hcsa,
the
picture
is
not
very
clear
how
we're
going
to
do
it.
Maybe
we
should
remove
it
as
a
feature
gate,
but
we
also
dropped
this
one,
and
this
was
very
late
in
the
cycle
like
how
do
we
in
a
more
granular
level,
control
what
timeouts
kubernetes
has
for
hcd
for
coblet
bootstrap?
A
A
We
did
a
lot
of
code
changes
all
the
prs
merges
at
this
point.
We
also
did
a
lot
of
debugging
and
getting
the
endpoint
test
to
be
good.
We
are
not
adding
documentation
for
this
feature
because
it's
alpha,
that's
like
standard
practicing,
cube
adm,
but
somebody
walked
an
issue
with
the
kubernetes
website.
A
I
don't
think
it's
linked
here.
So
basically,
the
idea
was
to
start
documenting
the
kubernetes,
the
kubernetes
feature
gates.
Even
if
it's
minimal,
we
should
have
some
sort
of
a
page
to
list
what
feature
gates
we
have.
Let
me,
let
me
open
a
new
window
to
show
you
what
I
mean.
A
So
core
kubernetes
currently
has
a
list
of
feature
gates
and
though
you
know,
those
are
not
ideally
documented
currently,
but
there's
still
some
enumeration
at
the
website.
That's
something
that
the
kubernetes
website
maintainers
really
like,
but
the
same
people
pointed
out
that
we
don't
have
one
for
cuba
at
the
bottom.
Here
I
think
there's
some
docks
explaining
what
the
future
gates
are
doing
and
we
don't
even
have
that
many
features
against
incubation.
Currently
we
have
only
two,
but
they
insisted
that
we
have
such
a
page.
I
did
not
disagree.
A
But
yeah
we
have.
This
is
our
band
for
122,
but
eventually
we
are
going
to
start
documenting
the
fissure
gates,
like
no
dogs
are
present
in
20
planet.
A
For
this
feature,
we
are
basically
going
to
look
for
feedback
from
consumers
and
eventually
decide
what
we
are
going
to
do
about
configuring
it,
because
the
feature
gate
currently
will
just
enable
it
always
by
default,
and
maybe
we
want
to
have
a
way
to
disable
this
or
make
it
like
somehow
configurable
for
particular
users,
maybe
in
the
next
cube
adm
api,
which
is
potentially,
if
you
want
beta4.
A
And
these
are
the
new
features
I
wanted
to
show
the
release
timeline
as
well
so
july,
8
conferences
tomorrow,
we
don't
have
any
critical
pending
changes
to
merge
until
tomorrow.
Test
freeze
is
is
also
this
related
to
tests
for
new
features.
We
already
added
our
tests.
A
A
Yeah,
so
I'm
hoping
that
somebody
will
help
us
with
that.
It's
going
to
be
fairly
easy
to
update
at
this
point.
A
A
Help
wanted
documentation,
fixes
and
maybe
open
discussion
on
these
things
that
we
couldn't
merge,
but
I'm
going
to
everything
that
is
design
and
feature
I'm
going
to
move
to
the
next
milestone
at
this
point.
Sorry,
after
code
for
this
tomorrow
and
yeah,
that's
what
we
have
it's
in
a
pretty
good
shape
this
release.
A
This
is
a
like
a
ps8
series
being
updated
to
3.5,
because
currently
we
are
using
a
rc
version
in
kubernetes,
which
is
it's
a
bit
absurd.
If
we're
going
to
release
122
with
a
pinned
version,
that
is
an
rc
so
basically
waiting
we're
currently
waiting
on
people
to
approve
the
changes
and
maintainers
of
the
remaining
area
that
has
to
see
approval
and
yeah.
That's.
A
So
that's
that's
all
I
had
do
you
have
any
questions
about
this
kubernetes
update
through
122.
A
Yeah
I
echo
your
shout
out.
I
already
gave
a
shout
out
to
the
google
contributor
in
the
shout
outs
shadow,
the
our
regular
chinese
contributors
have
been
doing
tremendous
work
to
help
us
with
all
of
these
features
with
review,
especially
the
the
v1
beta
3
api
tasks.
I
don't
think
I
could
have
completed
the
work
without
them
and
they
have
put
we
helped
with
review
and
prs.
So
that
was
great
collaboration.
B
Yeah,
it's
not
super
urgent,
it's
something
that
we
can
review.
Is
that
another
topic
in
an
agenda
or
no
other
things
that
you
want
to
talk
about,
but
if,
if
I
was
basically
checking
a
bit
of
code
of
kubernetes,
while
helping
colleagues
that
is
working
on
on
the
kubernetes
authentication
proposal
for
cluster
api
and
I
established
on
on
these
on
this
function
and
I'm
wondering
if
there
is
something
that
can
be
improved.
B
So
if
you
have
time
we
we
ca,
we
take
a
look
and-
and-
and
we
discuss
here
otherwise
we
can.
I
can
open
an
issue
or
we
can
have
a
separate
discussion
up
to
you
to
to
decide.
It
is
work
to
be
discussed
in
this
meeting
or
not.
B
A
B
B
B
A
B
A
B
I
think
that
the
check
should
be
the
really
the
first
thing
in
this
procedure.
So,
let's
see
if
there
is
already
another
node,
if
not,
then
I
have
to
do
to
to
do
whatever
is
necessary
to
start
on
the
node.
B
A
Well,
it's
if
it
fails.
It
fails.
Okay,
basically,
basically
you're
asking
that
we
shouldn't
write
the
bootstrap
config
to
disk
before
we
do
this
check.
B
A
I
see
we
also
apparently
fetch
the
ca
from
this
from
the
bootstrap.
This
has
to
happen.
B
A
B
Okay,
I
I
will
open
an
issue,
so
we
can
discuss
because
yeah
I
was
reviewing
this
code
with
a
colleague
and
this
sequence
kind
of
ringed
some
belt
to
me,
because
it
is
in
case
of
problem.
Basically,
we
leave
the
the
kubernetes
with
a
booster
configure
and
without
a
configuration
and
the
kubernetes
will
start.
I
guess.
B
Yeah,
but
but
as
soon
as
you
have
the
booster
configured,
the
kubrick
will
start
talking
to
the
api
server
yeah.
It
can
potentially
spawn
a
little
yeah,
and
so
I
want
to
be
sure
that
the
kubernetes
start
talking
with
the
api
server
when
he
gets
a
proper
configuration,
and
we
are
sure
that
we
have
to
complete
the
the
unit.
A
B
A
B
A
B
Yeah
yeah,
I
I
I
was,
I
was
discussing
the
kubernetes
authentication
for
copy
with
yuvaraji
and
we
were
investigating
how
it
works
just
half
an
hour
ago,
and
I
stumbled
into
this
procedure
and
while
basically
reading
this
procedure
with
yuvaraji,
I
noticed
that
the
order
your
thing,
the
ordering
of
things,
is
kind
of
confusing
to
me,
and
so
this
is
why
I
raised
this
point
in
here.
That's
it!
I
will
open
an
issue
by
not
worry,
so
we
can
discuss
and
take
the
proper
time
to
decide
if
to
go
on
or
not.
A
Michael
housler
from
aws,
I
think
I
explained
what
they
were
doing.
They
were
doing
something
like
a
mechanism
that
runs
in
parallel
to
cuba.
Adm
join
and
prepares
like
a
service
that
is,
in
the
background
that
controls
like
a
separate
authenticator
and
things
like
that.
A
It's
rather
complicated,
but
they
basically
wanted
to
for
us
to
provide
sub
phases
for
this
particular
function,
to
break
it
down
into
suitcases
which
in
a
way,
makes
sense,
but
a
couple
of
points
here
that
if
we
did
that
the
refactor
that
you're
proposing
will
be
a
breaking
change,
and
also
quite
unfortunately,
but
the
way
we
implement
these
phases
every
time
we
break
something
into
surpasses.
That
is
a
breaking
change
in
itself
because
of
the
all
argument.
A
At
the
end,
maybe
you
remember
the
discussions
with
rusty
and
rafael
and
the
way
we
did
it
is
every
time
you
break
something
into
phases.
It's
a
breaking
chain
which
is
difficult.
B
B
I
think
that
that
that's
the
the
request,
because
they
want
to
use
hardware,
verification
and
stuff
like
that,
and
it
is
now
difficult
to
to
plug
such
tools
in
into
the
kubernetes
flow.
That's
it
so.
A
B
A
All
right,
if
you
add
the
help,
wanted
label,
I'm
sure
that
somebody
will
like
to
refactor
it
and
if
you
provide
sufficient
details
like
with
blobs
of
how
to
move
things
around
them,
somebody's
going
to
take
it
for
sure.
A
Okay,
okay,
I
don't
think
I
have
anything
else.
Scott,
do
you
have
any
topics
for
today,
nothing
from
my
side.
A
Thank
you
for
joining
and
see
you
again
in
a
couple
weeks.
Bye.