►
From YouTube: SIG Cluster Lifecycle 2020-08-25
A
A
A
A
A
A
A
A
A
A
But
the
core
of
the
problem
is
that
the
wall
of
the
six
don't
invest.
You
know,
effort,
don't
invest
human
power
into
the
flaking
and
everything
goes
to
the
release,
team
ci
signal
and
they
start
being
in
the
six
and
then
the
six
dollar
response,
and
it's
we
have
a
very
unstable
release
in
terms
of
ci.
A
So
hopefully
this
is
going
to
change
the
picture.
If
you
have
prs
in
kubernetes
kubernetes
note
that
the
peers
are
now
supposed
to
merge
only
by
milestone
maintainers,
applying
the
milestone,
120
label.
Sorry,
the
mouse
play
the
milestone,
and
only
after
that,
the
pr
can
merge,
but
also
the
the
bowser
maintainers
have
to
be.
A
You
know
they
have
to
follow
this
rule
that
low
priority
appears
first
features
last,
but
if,
like
I
don't
see
what
happens,
if
you
have
a
good
cleanup
here
that
comes
like
very
late
into
the
release
in
between
the
feature
appears
a
response
budget.
Are
you
supposed
to
hold
it
because
it
was
not
in
the
window
where
low
priority
stuff
is
supposed,
but
you
know
there's
some
lack
of
clarity
around
these
instances,
but
yeah.
I
just
wanted
to
give
this
psa.
A
B
Just
to
your
last
point
about
when
you
can
merge
like
a
cleanup,
I
don't
my
reading
just
reading
this
document
is
that
the
the
steps
are
not
exclusive
I.e.
You
can
always
merge
a
fix
that
appear
that
fixes
failing
tests,
even
if
we're
in
the
regular
maintenance
step.
So
you
can
merge
from
any
prior
step
is
my
is
my
reading,
but
it's
not
like
it's
not
like.
They
are
if
you
missed
the
window,
you're
you're
you're
done
and
for
like
a
while.
B
C
D
A
A
I
by
my
comment
of
this
on
the
original
document,
was
that
maybe
we
should
just
in
the
future
baby.
We
should
open
master
right
away,
and
you
know
have
this
stable
branch
on
the
side,
open
master
right
away,
but
the
problem
with
that
is
that
it
requires
more
people
to
watch.
Both
the
stable
branch
hit
the
bus
to
develop
it
to
make
sure
that
nothing
breaks.
A
A
Right
moving
to
the
next
topic,
the
war
group
naming
email
that
I
said
did
not
get
any
responses.
I
think
I
said
it's
not
like
10
days
ago,
but
I
saw
that
a
lot
of
topics
were
created
right
after
my
email,
so
maybe
it
spawned
some
ideas
for
the
working
group
such
as
hey.
We
want
to
work
on
the
list
of
words
that
are
offensive.
A
I
saw
a
lot
of
threads
spoiled
after
that.
So,
if
you're
interested
in
these
discussions,
please
watch
this
mailing
list.
For
instance,
I
see
a
discussion
between
jordan
and
some
of
the
working
group
creators
around
the
world
aboard
because
abort
apparently
is
it
has
a
dual
meaning.
It's
considered
offensive
with
respect
to
females,
and
if
you
have
this
word
and
if
they
decide
that
the
board
is
of
s,
if
you
have
to
change
it,
it's
probably
gonna
be
a
breaking
change.
A
I
think
that
jordan
might
be
objecting
here
in
his
last
comment:
yeah
yeah,
so
the
list
of
words
is
not
clear.
Yet
first,
those
have
black
box.
This
is
more
of
an
expression.
I
guess,
but
once
we
have
this
list,
I'm
going
to
assume
that
it's
going
to
be
very
controversial
and
also
it
might
be
very
breaking
to
some
of
the
codes
that
we
have
in
the
sub
projects
of
this
sick.
A
So
yeah,
please,
please
watch
the
discussions
once
we
have
this
list.
Of
course,
I'm
going
to
again
summarize
everything
and
you
know,
link
to
an
official
document
that
enumerates
the
problem
words.
We
didn't
get
a
response
about
the
timeline
which
tells
me
that
they
don't
have
a
timer
timeline
yet
so
I
don't
think
they
they
have
an
idea
how
long
this
is
going
to
take.
Is
it
the
2021
schedule?
Is
it
a
very,
very
long
term
effort?
A
A
B
B
Rather
than
saying
like
I
like
what
you
did
with
c
cluster
lifecycle,
I,
like
spiff
xps,
like
what
should
we
call
the
branch
formerly
known
as
master
like,
I
feel,
like
that's
a
very
productive
thing
to
do
like
what
should
we
call
leader
elect?
Are
we
okay
with
leader
election
right
or
whatever?
It
is
like?
You
know,
like
very
specific
examples.
B
The
the
the
idea
of
coming
up
with
lists
of
languages
is
certainly
we
should
highlight
the
things
that
we
consider
to
be
problematic,
but
the
general
blanket
translations
are
going
to
be
difficult
to
do
in
the
abstract.
I
think
so.
I
think
if
we
can,
if
we
can
keep
nudging
them
towards
specifics,
I
think
that
will
help
advance
this
working
group.
D
D
B
A
Yeah,
I
think
that
they
could
define
the
list
of
offensive
terminology,
and
maybe
because
I
see
some
not
so
good
attempts
to
define
alternatives,
maybe
just
they
should
ask
the
developers.
I
guess
the
engineers
should
come
up
with
some
of
the
alternatives,
because
it's
oftentimes
it
depends
on
the
context
like
it's
yeah.
A
A
All
right
so
today
I
checked
the
survey.
We
have
67
responses,
thus
far,
I'm
kind
of
surprised.
The
number
is
not
that
high
yet
compared
to
the
cuba,
the
app
survey,
I
think
some
folks
on
this
meeting
should
have
added
access.
You
can.
You
should
be
able
to
view
the
responses,
it's
not
for
the
rest,
but
I'm
proposing
that
maybe
we
should
expose
the
results
on
the
september.
A
F
A
A
At
some
point,
I
tried
using
reddit
and
I
think
they
deleted
my
post,
because
I
think
it's
against
their
policy
of
I
think
they
they
have
some
sort
of
bots
that
deletes
surveys
taking.
There
is
some
sort
of
spam,
so
I'm
not
going
to
share
it
on
reddit,
but
you
know,
if
you
have
other
ideas,
please
share
it
further.
A
And
some
of
the
responses
are
very
interesting,
especially
around
docker,
because
you
know
signal.
Has
this
idea
to
remove
docker
support,
remove
docker
ship,
but
the
award
of
people
docker
still,
and
I'm
going
to
share
this
this
specific
question
with
them
to
see
the
number
of
people
responding
to
this,
how
they
responded,
and
maybe
we
should
start
thinking
how
we
can
move
this
into
community
slash
user
maintenance
instead
of
removing
it
completely
yeah,
so
yeah.
This
is
going
to
be
very
useful
for
us.
A
A
A
C
C
We
improved
the
externalities
of
the
support
in
the
government
control
plan
and
also
we
addressed
also
some
feedback
from
from
from
kuberconf
and
and
basically
we
are
simplifying
the
cup
of
the
quick
start
now
cut
these
you.
You
can
basically
make
the
the
quick
start
with
cup
d,
like
any
other
provider,
so
on.
No,
I
can
require
the
artifact
will
be
published
and
so
on,
but
there
are.
C
There
are
plenty
of
warning
telling
that
company
is
not
from
production
in
a
use
case,
if
a
production
use
case
yeah
and
also
we
improved
the
docs
around
the
feedback
from
kubercon,
so
kubercon
user.
After
listening
about
customer
bi,
they
were
testing
it
and
they
found
some
pieces
of
of
the
docker
kind
of
a
hard
to
understand.
So
we
addressed
some
comments
in
order
to
make
things
easier
for
the
newcomers
and
yeah.
I
I
give
also
a
quick
update
on
occupy
the
07.1
releases
out.
A
A
C
A
C
A
C
A
C
So
the
domain
use
case
is
for
is
is
for
newcomers
that
want
to
give
a
try
to
cluster
api,
but
they
don't
have.
They
don't
want
to
to
use
a
real
cloud
provider
and
also
for
for
testing
and
basically
before
we
were,
we
were
kind
of
of
keeping
in
separated
the
the
top
of
the
provider
from
for
the
from
the
others,
because
it
is
not
for
production,
but
at
the
end
the
user
were
complaining.
D
A
C
C
The
kappa
v
basically
started
to
deploy
a
new
release
of
of
ccm
and
the
new
release
of
ccm
included,
the
apr
that
basically
introduced
a
regression
which
is
now
being
discussed
if
it
is
a
problem
of
how,
on
which
side
that
basically,
the
problem
is
so
there
is
discussion
around
this,
but
the
there
was,
since
there
is
a
flag
that
allows
to
make
ccm
to
work
with
the
or
the
the
for
the
behavior
minus
one.
We
are
setting
this
behavior
while
we
we
understand
the
problem.
A
B
Jesse
cops,
I
think
yes,
not
time
to
report
basically
in
a
sort
of
quiet
period
in
terms
of
releases
and
bug,
fixes
we're
basically
more
or
less
caught
up
and
actually
we're
working
on.
I
guess
bigger
features
for
the
future
and
I
think
the
most
interesting
one
is
like
what
we're
calling
like
a
cloud
integrated,
node
authentication
so
using.
B
This
assumes
that
whoever
is
actually
doing
the
authentication
has
access
to
sign
those
those
certificates
and
that
again
can
be
done
using
like
okay.
What's
it
called
the
key
kms,
the
key
management
service
on
aws
or
kms
on
gcp
as
well,
so
it
it,
but
that
it
would
be
a
more
powerful
or
privileged
machine
that
would
have
to
be
able
to
answer
that.
B
Oh
and
the
other
one
we're
working
on
is
the
jwks
getting
that
integrated
in
so
that
already,
because
we
need
to
get
that
that
functionality
into,
I
think
beta.
I
think
it's
an
alpha
right
now.
We
need
to
get
into
beta,
and
then
that
requires
test
coverage,
so
we're
trying
to
get
some
test
coverage
on
that
functionality.
What
that
will
allow
is
it
will
allow
you
to
use
your
kubernetes
service
accounts
to
authenticate
to
aws,
for
example,
and
almost
gte
or
gcp,
but
definitely
deadps.
A
B
I
mean
that's,
that's
the
hope
and
we'll
we'll
see
if
it's
good
and
honestly,
I
don't
know
where,
like
that's,
because
it
works
great
in
cops
where,
where
do
we
put
this
like?
How
do
we
get
it
so
that
everyone
can
use
it?
I
honestly,
I
don't
know
where
it
wherever
it
goes,
but
we
can
cross
that
bridge
later,
let's
first
figure
out,
if
it's
any
good
and
then
yeah,
it's
not
clear
whether
it
goes
into
the
aws
cloud
provider
project,
it's
not
clear
whether
it
goes
into
something
a
new
oauth
component.
B
B
A
B
I'll
look
at
what
we're
doing-
I'm
not
I'm
not.
I
I'm
not
doing
most
of
the
work
on
this,
so
I'm
entering
and
work
on
this.
So
I'm
doing
the
review
what
I'm
reviewing
but
not
really
doing
much
of
the
work.
So
I
will
review
exactly
what
we're
doing
there.
I
can
report
back
next
time
but
yeah.
I
agree
that,
like
you
know,
we
we
would
like
to
have
not
have.
Every
team
have
to
reinvent
the
wheel,
particularly
on
what
is
a
very
security,
sensitive
area.
A
I
think
it's
this
bike
bike
that
is
who
is
driving
this
on
your
side.
I
believe
they
had
some
sort
of
a
meeting
with
you
know.
Duffy
cooley
was
there
nadir
from
where
a
lot
of
different
companies
met
at
this
secret
meeting.
I
believe
that
call
north
america
last
year
and
they
discussed
this
not
identity
problem
like
apparently,
I
believe
google
are
one
of
the
few
providers
that
have
a
solution
for
it,
but
it
it
was.
You
cannot
integrate
it
with
other.
A
I
you
know,
I
don't
know
the
details
there,
but
it's
a
kubernetes
problem
in
general,
like
nobody
can
sign
this.
The
coup
control
manager
cannot
sign
this
certificate,
so
people
have
to
integrate
their
own
controller
and
apparently
some
people
already
did
cuba
dm
is
using
a
self-site
service
certificate
for
the
couplet,
and
somebody
created
this
signing
controller
that
they
they
can
actually
approve
the
csrs
that
for
the
service
certificate,
but
what
happens
there
that
somebody
discovered
that
this
controller
is
not
secure
and
nobody
is
fixing
the
control.
B
B
A
Yeah
I
mean
this
is
not
asking.
I
just
tried
to
comment
to
this
yeah.
I
think
that
sea
golf
should
be
responsible
for
creative
potential,
this
working
group-
if
somebody
wants
to
invest
into
it,
they
have
a
unfortunately,
of
course,
due
to
limited
bubble.
If
sea
golf
have
a
lot
of
undocumented
topics,
I
would
say
recently
we
documented
ca
rotation
like
nobody
in
the
public
knew
how
to
rotate
ca
properly
kubernetes.
A
A
For
instance,
something
else
that
is
not
documented
is
what
are
we
doing
with
some
of
those
flags
for
the
kcm
that
don't
support
bundles,
like
when
you're
rotating
a
ca?
Supposedly
you
have.
You
have
to
have
support
for
bundles
some
of
those
flags
that,
for
instance,
that
side
the
class
that
verify
the
class
certificate
authorization
they
sorry
authentication.
They
do
not
support
buttons.
A
G
A
A
C
A
A
So
you
know
my
marketing
person
is
asking
me
hey
how
many
people
watch
your
talk
and
I
don't
know
I
can
only
say
how
many
people
register
for
the
talk
like
we
had
jessie
had
something
like
96,
I
believe,
but
maybe
you
know
usually
it's
much
lower
than
that.
So
maybe
30
people,
but
I
cannot
say
the
number
because
I
don't
know-
and
this
is
something
that
I
gave
you
the
feedback.
I
I'm
not
sure
if
everybody
completed
the
called
survey
after
the
event,
but
they
had
a
survey.
A
A
I
think
the
deadline,
for
that
is
the
18th
of
september.
I
guess
in
the
next
sequence,
live
psychometric
we
can
discuss
which
talks
we
can
have
at
cubecon
na
and
from
the
response
of
nancy.
She
said
that
we're
doing
the
same
thing
as
europe.
My
question
was
very
specific:
can
you
give
us
one
sick,
sick
session
plus
three
subproject
sessions
and
she
just
replied
that
we
are
doing
the
same
thing,
which
maybe
means
that
we
can
do
the
same
number
of
sessions?
I
don't
know.
A
B
I
have
one
question:
I
don't
know
lubria
the
talk
that
you
and
I
gave
was
we
had
no
real
audience
interaction.
The
audience
was
able
to
send
in
questions
via
chat.
It
was
not
high
bandwidth
communication,
it
was
not.
I
don't
know
if
anyone
experienced
a
different
platform
that
would
be.
You
know
like
I,
at
the
last
kubecon
north
america.
B
I
very
much
enjoyed
doing
a
like
round
table
over
lunch
where
there
was
very
much
like
you
know,
not
not
a
talk,
but
just
like
interaction
with
the
community
and
everyone
getting
together
and
talking,
and
I
thought
that
very
valuable,
but
obviously
that
would
not
work
in
the
format
we
you
you
and
I
experienced
the
premiere
where,
like
one
person
was
had
video
and
everyone
else
was
texting,
I
don't
know
if
anyone
had
a
better
experience
or
a
different
experience.
That
would
be
more
amenable
to
by
rational
communication.
A
Okay,
yeah,
it
was
just
another
like
like
a
user
developer
type
of
communications
slack,
which
we,
you
know,
we
have
this
every
day,
pretty
much
so
the
maintainer
track
was
definitely
lacking,
that
there
was
no
maintainer
track
at
this
cuckoo.
I
really
enjoyed
like
just
in
the
the
same
way
you
were
saying
on
the
table
with
people
having
lunch.
You
know
we
have
discussions
like
with
tips
and
care
lucas
sitting
on
the
table,
having
chat
about
important
things,
and
this
is
like
it's
not
the
same.
H
H
H
H
A
Yeah
yeah.
Well,
that's
why
they
requested
that
people
joined
30
minutes
earlier.
I
you
know,
one
person
is
telling
me
adjust
your
camera
brightness,
I'm
saying
hey!
This
camera
cost
10
dollars.
I
cannot
you
know
if
I
touch
it.
It's
going
to
break
apart
like
it's
horrible.
The
other
person
is
telling
me
your
microphone
is
quality
is
not
that
good.
Like
you
know,
my
microphone
is
even
cheaper
than
that.
A
E
B
A
Yeah
somebody,
I
think,
josh
burkus
from
red
hat,
had
this
idea
that
we,
you
know
after
cube
code,
we
can
have
these
round
tables,
but
in
virtual
zoom
and
somebody's
moderating
people
raising
hands,
they
can
rotate
on
the
topic.
But
I
get
like
what
is
what
is
the?
What
is
the
agenda
of
the
specific
topic.
C
For
instance,
I
I
participated
at
the
mentoring
session,
mentoring
and
mentor,
and
this
was
really
well
organized
because
there
was
a
platform
with
many
beautiful
tables
and
people
were
when
they
joined.
They
basically
click
on
a
seat
around
a
table
and
the
mentor
were
moving
after
some
slot
from
one
table
to
another
and
and
in
and
basically
every
table
was
composed
by
seven
people,
the
mentor
and
sixth
mentoree
and
each
me
so
that
there
was
chance
to
do
introduction
to
to
answer
a
question
for
each
person.
A
Yeah,
I
think
this
was
the
idea
that
josh
pretty
much
also
suggested
it
has
to
be
small,
the
number
of
people
in
the
the
room,
but
you
know
there
was
no
accidental
maintainer.
Sorry,
I
guess
how
we
track
in
a
way
this,
though
you
you're
not
assigned
randomly
to
talk
to
people
like
it's
yeah,
it's
definitely
different,
but
maybe
for
north
america
they're
going
to
think
of
some
sort
of
a
solution.
A
Then
again,
north
america
is
kind
of
questionable
so
early
after
europe,
because
it's
the
same
culture
like
I
don't
think
we
could
change
that
much
in
our
sub
projects
in
our
sick
intro
between
this
keep
calling
the
next
good
call.
So
I
guess
we
should
just
submit
and
repeat
the
whole
process.
It's
kind
of.