►
From YouTube: 20191009 - Cluster API Office Hours
Description
20191009 - Cluster API Office Hours
A
Hello
today
is
third,
our
Wednesday
October
9th.
This
is
the
cluster
API
office
hours.
Meeting
cluster
API
is
a
project
of
sit
closer
lifecycle.
If
you
are
new
or
need
a
refresher,
we
do
have
meeting
etiquette.
So
please
try
to
use
the
rave
hand
feature
on
zoom'
and
I
will
do
my
best
to
make
note
and
call
on
you.
If
you
have
agenda
items,
please
go
down
below
in
this
document
and
add
to
the
agenda.
It
looks
like
we're
pretty
light
today
and
please
also
add
your
name
to
the
attending
list
as
well.
A
B
C
A
F
C
So
the
focus
on
the
PC
was
the
day
one
experience
so
today,
I'm
going
to
show
how
it
looks
of
the
video
it
will
look
like
and
give
a
brief
explanation
about.
How
do
you
plan
to
achieve
DISA
and
hopefully
also
answer
two
questions?
So
we
forget
starting
one
of
the
cornerstone
of
the
original
proposal
about
Custer
cattle
orchestra.
Red
bean
was
to
assume
that
the
user
will
provide
a
kubernetes
cluster
by
their
own,
and
this
this
makes
a
little
bit
different,
the
workflow.
So
even
a
fact,
the
fact
that
kubernetes
cluster
already
exists
v.
C
The
first
step
for
cluster
cluster
will
be
to
initialize
the
management
cluster
by
standing
the
cluster
API.
So
there
is
no
more
provisioning
with
mini
group
of
cooperate.
Quick
look
at
at
the
cluster
cutter
little
init
command
here.
The
most
interesting
thing
is
that
you
can
specify
which,
which
or
which
booster
provider
you
want
to
Eastern
the
faulty
stupid
min
and
the
same
in.
At
the
same
way,
you
can
specific
that
you
can
specify
one
or
more
infrastructure
provider
that
we
want
to.
Mr.,
so
they
made
the
LDR
is
ear.
C
C
C
What
I
think
that
is
more
interesting
for
for
this
audience
today
is
a
little
bit
deep
dive
of
about
all
this
is
going
to
a
pen.
So
the
first
important
thing
to
know
is
that
cluster
cutter,
as
least
as
a
predefined
list
of
provider,
for
instance
the
least
say
saying
copies-
is
a
core
provider,
and
this
is
the
kubernetes
repo
where
the
core
provided
is
available.
I
want
to
order
list
and
latest,
and
this
is
the
file
that
I
have
to
to
grab
in
order
to
do
the
installation.
C
Another
thing
that
can
be
interesting
is
that
is
looking
at
basically,
water
expectation
or
vironment.
Custer
cattle
is
expecting
for
the
providers
and
we
try
to
keep
them
really
simple.
So
this
is
my
free
kappa
provider
and,
as
you
can
see,
there
is
exactly
the
same
infrastructure
provider
that
is
file
infrastructure
component
file
that
you
can
find
extrema,
so
it
can
contain.
C
It
might
contain
valuable
substitution
and
castor
cattle
with
takes
care
of
doing
variable
substitution,
but
also
there
are
these
files
that
are
the
template
that
are
used
when
that
were
used
when
I
be
the
pastor
cathode
create
this
files
are
in
fact
real
template
here,
I
have
one
already
open,
and
so
they
are
gota
plate.
You
can
do
variable
and
castle
cattle
is
providing
a
set
of
well-known
variable
that
you
can
use
to
customize
your
to
plate.
C
Create
different
on
different
number
of
nodes
and
so
on.
So
this
is
the
the
biggest
requirement
that
custer
cutter
is
giving
to
provider.
That
is,
okay.
We
are
provided
if
you
want
the
that
caustic
cattle
being
able
to
create
a
template
for
you,
the
provider
have
to
create
a
template
one
or
more.
This
is
a
different
template
that
I
and
the
template
has
to
follow.
C
C
The
last
bit
that
I
want
to
show
you
is
is
that
we
are
considering
to
also
to
add
a
new
sir
D
managed
by
cluster
cutter.
That
allows
to
keep
some
metadata.
Some
information
about
the
distal,
a
provider
like
the
version
number
or
mine
might
be
they
watch.
The
namespace
provider
is
watching
that
this
information
could
be
useful
again
for
managing
up
a
grade
or
for
giving
warnings
to
the
user
when
the
user
is
is
doing
things
that
can
compromise
the
that
might
compromise
the
management
cluster
like
standing.
A
G
Actually
that
clarify
a
little
bit
later,
I'm
sorry
for
that,
my
breasts
entered
too
soon
so
there.
The
idea
here
is
I
think
that
the
answer
is
actually
the
coop
config.
So
for
briefs.
You
think
about
the
the
cluster
where
you
install
cluster
API
and
everything
else.
I
was
thinking
immediately
when
I,
when
I,
when
you
start
presenting
this
demonstration
that
cluster
control
would
actually
be
self
deployable,
but
I
think
it's
not
the
case
right.
G
C
G
C
A
Bootstrapping,
how
does
this
work
in
situations
where
data
is
generally
based
on
other
generated
data?
The
push
to
the
cube,
ATM
config
resource
means
that
the
generators
and
the
infrastructure
providers
are
sometimes
rather
involved.
It
seems
like
it's
satisfies
Kappa,
but
it
won't
work
with
cat
B
as
it
is
today
due
to
the
fact
that
cube
ATM,
config
resource
must
know
everything
in
advance,
I,
don't
think
this'll
work
for
cat,
B
or
possibly
others
until
there's
the
ability
for
the
infrastructure
provider
to
modify
the
config
resource
at
runtime.
So
what.
C
C
H
I
get
that
that's
why
I
said
I
think
it's
related
daniel's,
whether
it's
the
assets
that
need
to
be
created
for
AWS
and
that
happens
via
some
workflow
outside
the
scope
of
this
pretty
nice.
You
know
refactor
question
cuddles
work
on
that,
but
there
seems
to
be
provider
specific
workload
that
might
be
related
to
how
the
manifests
or
generated
the
data
that
goes
into
them.
It
might
be
related
to
non
manifest
assets,
but
right
now
it
seems
restricted
to
interpolation
environment
variables,
and
maybe
there
needs
to
be
the
that.
H
Could
I
made
a
recommendation
in
chat
for
it,
so
you
could
take
a
look.
There
suggested
go
templates
for
more.
You
know
advanced
way
to
do
that,
but
it
might
also
need
to
include,
like
perb
provider
infrastructure
provider.
In
fact,
look
here's
is
the
plug-in
or
here's
executable
that
it
will
run
as
part
of
this
and
dump
assets
to
a
known
directory
or
or
perform
some
behavior
I.
Think.
I
C
I
H
Is
there
is
a
use
case
with
cap
v,
it's
where
if
a
resource
is
created
that
would
have
otherwise,
then
and
by
the
way
that
created
resource,
has
some
identifying
piece
of
information
or
they're
just
text
whatever,
but
that
has
to
be
part
of
the
cube
ATM
config
resource,
because
at
runtime
the
infrastructure
providers
can
no
longer
modify
that
that
has
to
be
submitted.
As
is
then,
there
may
be
a
use
case
where-
and
there
is
today
where
something
has
to
happen
before
that
queue.
Midium
config
resource
is
created.
So
it's
created
with
that
data.
J
I
just
want
to
be
interjected
like
there's
a
proposal
out
and
like,
and
this
is
like
I
don't
know
mind.
This
is
just
a
POC-
doesn't
have
to
solve
all
the
problems
today
and
we
can
take
definitely
that
into
account
in
the
proposal
and
yeah
I
started
doing
that
this
morning.
So
I
would
suggest
to
just
move
the
conversation
there
so
that
we
don't
go
too
much
into
implementation
details
right
now.
A
K
Hey
everyone,
so
just
an
update.
The
upgrade
tool
is
now
fully
upgraded
to
work
against
V
1,
alpha
2
on
master.
So
definitely
you
feel
free
to
play
around
with
it
test
out
and
file
bugs
so
Andrew
Rideau
is
an
awesome
and
has
been
testing
it
testing
the
tool
out
on
his
multi
node
control
planes
and
it's
definitely
buggy
for
multi
node
control
clans.
So
this
is.
This
is
being
tested
in
that
environment
as
well.
K
A
Okay,
before
I
go
on
to
backlog
grooming,
does
anybody
have
anything
else
that
they
want
to
talk
about?
We
actually,
we
probably
should
check
in
on
the
caps
so
before
I
get
into
backlog
grooming.
Let's
do
that
the
machine
pool
one
Cecille
I
see
your
name
here
in
my
distant
list.
Do
you
have
any
updates
on
an
ETA
for
when
this
is
ready
for
comments?
A
A
Yeah
so
I,
okay,
so
I
have
I'm
in
the
process
of
updating
the
document.
With
you
know
some
of
some
of
the
feedback
that
I
got
with
the
cap
see
working
group,
including
because
Liz
and
Steven
Augustus
and
company,
so
I
should
have
that
document
updated
by
the
end
of
end
of
day.
Today,
so
I
think
people
can
start
to
look
at
that
doc
and
it'd
be
nice
to
start
to
have
some
feedback,
some
comments
there
and
then
pretty
shortly
after
I,
should
have
my
POC
ready
for
you
know
for
demo
and.
A
A
L
Sorry
I
couldn't
find
a
mute
button,
so
we
had
our
first
kickoff
meeting
I'm
back
on
the
first.
We
are
thinking
about
having
another
one.
Now
that
we
have
had
some
feedback
on
our
kept
document.
I
think
we
said
earlier
this
week
or
maybe
even
earlier
today,
that
we
were
gonna
have
that
ready
by
end
of
week
for
everyone
to
review.
A
Okay,
testing,
yeah
I've.
M
M
A
Thank
you
and
no
remediation.
Do
we
have
you,
don't
see
Michael
or
min
all
Michael
rhythmic
and
are
you
in
the
loop
at
all
on
Notre
mediation
with
the
other
Michael
or
separate
I'm,
not
up
to
speed
on
the
latest?
On
that
all
right,
no
worries.
Thank
you
and.
A
B
K
H
K
K
H
Henry,
yes,
Amy
again,
thank
you
for
the
cluster
up
here
to
live
actually
literally
been
waiting
on
that
I.
In
the
meantime,
my
backported
part
of
the
controller
runtime
twos
client,
because
I
couldn't
know
crate
to
the
I
couldn't
because
of
that
brizel
I
just
checked,
so
it's
only
a
master
at
this
time.
Did
you
already
address
that?
Is
there
going
to
be
a
tag
later
today
or
this
week?
H
H
A
H
A
C
The
idea
comes
from
the
word
compress
that
cut
all
the
while
it
can
try
to
think
how
to
implement
a
generic
mechanism
for
people
think
of
are
politically
figs
and
while
custard
cutter
can
take
responsibility
for
labeling,
they
provide
a
component.
The
idea
is
to
have
cluster
API
controllers
to
take
responsibility
to
label.
Everything
is
part
of
the
cluster
and
all
as
soon
as
we
have
everything
level.
Then
operation,
like
the
late
of
people
thing
will
be
much
more
easier.
A
A
A
The
next
issue
was
one
that
I
opened,
which
is
not
really
something
that
targets
a
specific
milestone,
but
it's
about
refining
the
release,
notes
process.
So,
to
give
an
example
of
why
I
created
this
issue,
we
had
a
change
that
we
made
to
the
leader
election
ID
for
the
controllers
that
are
in
cluster
API.
The
cube
am
bootstrap
provider,
Kappa
cap,
the
cap,
Z
cap
D
and
any
else
any
other
ones
that
we
could
find,
and
this
was
so
that
the
pods
could
run
together
simultaneously
in
a
single
namespace.
A
This
was
a
an
action
required
type
of
thing
for
the
release,
notes
and
the
tool
that
we're
currently
using
to
generate
release,
notes
what
it
does
is.
It
takes
a
look
at
people,
requests
and
looks
at
the
the
emoji
that
we
have
at
the
front
whether
it's
a
bug
or
sparkles
or
the
running
person,
and
essentially
takes
all
of
the
PRS
that
are
merged
as
in
a
release
and
generates
release,
notes
that
look
like
this,
but
it
doesn't
do
well
when
you
have
something
action
required
things.
A
So
the
information
that
we
have
here
about
the
leader
election
change,
that
I
just
made
was
information
that
we
manually
wrote,
and
so
the
reason
for
this
issue
is
to
try
and
find
a
way
to
allow
us
to
write
down
the
release,
notes
but
all
requests
if
possible
and
then
get
that
automated
into
the
github
text.
So
I
know
that
there's
some
tooling
from
tests
in
front
to
try
and
help
with
this
or
the
release
repo
to
help
with
this.
A
So
basically,
I
would
like
to
try
and
see
us
get
this
done
as
time
allows,
but
as
soon
as
possible.
So
I
think
putting
this
into
the
next
milestone,
makes
sense,
and
the
priority
I
would
say
is
important
soon.
I,
don't
know
if
if
anybody
has
any
strong
opinions,
otherwise,
but
that's
just
when
I'm
thinking.
F
Yes,
I,
don't
I
see
this
is
being
a
little
bit,
it
hasn't
been.
It
hasn't
been
too
difficult
to
manually
write
these
reliefs.
That's
the
additional
action
release,
notes
I'm
curious.
What
like
what
we
might
be
looking
for
out
of
this
like
it
is
it
is.
It
is
this
coming
from,
because
it's
too
onerous
to
keep
doing
this
by
hand
or.
M
So
I
think
my
problem
is:
is
that
it's
error-prone
right
now
and
that,
based
on
the
generated
release,
notes
unless
you
specifically
know
based
on
the
PR
number
or
the
title
that
you
have
to
provide
additional
context?
You
don't
know
that
so,
like
the
original
recent
copy
release
that
went
out,
didn't
have
the
action
item
and
we
had
to
go
back
and
add
that
to
the
notes
after
the
release
was
published.
So
andr
say
you
had
your
hand
up
briefly.
A
H
I
was
just
gonna
say
that
I've
been
waiting
for
this
for
a
while,
and
one
thing
I
do
just
maybe
it
will
help.
Others
is
in
yeah
and
I.
Guess
you
have
to
be
an
admin
on
the
repo
to
do
it,
but
you
could
also
just
do
it
in
a
local
app
I,
literally
just
keep
I'm
responsible
for
tallying,
and
if
you
are
for
your
project,
I
keep
a
running
like
markdown
file
of
what
that
really.
You
know
they
didn't
release
description.
H
A
And
Chuck
my
response
to
your
question
is
that
it's
always
easier
to
write
a
release.
Note
when
you
open
up
a
poor
request
than
to
remember
two
weeks
or
two
months
later,
that
oh
yeah
I,
there's
a
breaking
change
that
I
need
to
document
so
that
that
was
my
motivation
for
opening
this.
Not
that
it's
hard
to
edit
a
draft
of
the
release,
notes,
but
more
that,
if
there's
any
gap
in
time,
it
makes
it
a
lot
harder
to
remember
to
keep.
A
Okay
and
then
we
also
have
make
sure
that's
clear
and
for
the
cubm
bootstrapper,
let's
see
what
we've
got
one
excellent.
This
is
from
Daniel
and
I
know
there
was
some
back
and
forth
on
this.
Is
this
still
something
that
you
think
needs
to
go
in
Daniel,
based
on
the
summit,
Chuck
and
Jason's
comments.
A
B
Yeah
I'm
not
sure
that
the
idea
is
that
so
I
don't
think
it's
I,
don't
think
it's
actually
possible
today,
with
with
Kubb
ADM
in
its
current
state,
so
yeah
there
is,
there's
gonna
be
some
so
that
I
think
it's
blocked.
The
idea
is
that
you
know
we
specify
some
coup
BDM
configure.
We,
the
the
the
bootstrap
controller,
specify
some.
You
know
some
minimal
subset
of
the
Covidien
configuration
that
eventually
by
the
cube,
idiom
binary
right,
well,
values
will
be
defaulted
and
they
will
be
applied.
B
It
will
be
nice
as
as
an
operator
to
know
what
those
actual
values
are
and
currently
there's
there's
not
oh
there's
not
a
way
to
actually
get
at
at
those
values,
I
think
in
using
cube
ADM,
so
III
think
this
is.
This
is
a
similar.
So
if
anybody
is
familiar
with
the
couplet
configures,
the
end
point
I
think
the
the
problem
is
similar
right.
It's
I
want
to
know
sort
of
what
what
the
actual
configuration
is
and,
yes
I,
think
I
think
it's
a
good.
It's
a
good
feature
to
have.
B
A
B
Need
to
first
I
think
maybe
ask
some
of
the
comedian
folks,
you
know
what
what
what
the
work
involved
there
would
be,
but
yeah
I
mean
I
definitely
be
happy
to
to
Shepherd
this
too
long.
I,
don't
I,
don't
know
if
it's
going
to
happen
in
the
next
next
release,
though,
because
I
think
it
would
need
it
may
be.
An
additional
cuvee
DM
release.
A
Okay
does
next
seem
appropriate,
just
the
market
is
triaged.
Okay
and
I'm
gonna
go
with
long
term
sound
right
and
Chuck.
I
see
you
have
your
hand
up.
F
A
H
Yeah
I'm,
sorry
everybody
but
I
just
want
to
raise
the
issue
of
the
cube,
a
medium
bootstrapper
defaults.
Sort
of
it
seems
to
be
related
to
this
other
one,
but
I
think
you
said
that
we
had
somebody
that
could
work
on
it,
but
weren't
true,
it
was
good,
it
was
gonna,
be
ready.
I'm,
curious
I
can
do
anything
to
help
with
that.
It's
the
Jason
I
forget
the
issue
number,
but
it's
the
issue
around
providing
defaults
for
a
cluster
to
merge
the
man
with
the
bootstrapper.
H
M
H
H
Okay,
so
no
no,
it
wasn't
assigned
I
thought
we
must
be
thinking
another
one
but
yeah.
This
is
super
important
to
Academy
in
other
projects
to
be
able
to
have
cube
ad,
have
the
bootstrapper
say
Oh
for
this
provider.
There
are
some
sane
defaults
that
we
want
to
merge
in
if
these
things
are
empty
and
so
that
you
know,
operators
or
IT
admins
can
set
up
these
defaults
and
not
require
a
user
generating
the
manifests
and
to
know
everything
that
they
need
to
know
like
custom
repositories
or
extra
files
to
add.
A
So
we've
talked
a
little
bit
about
trying
to
find
a
way
to
have
sensible
defaults.
If
you
don't
want
to
specify
anything
at
all,
so
this
would
be
like
a
day.
One
kicking
the
tires,
make
things
easier
type
of
experience
that
also
could
apply
beyond
that
as
well.
So
I
don't
know
that
this
matches
exactly
with
what
you're
asking
for
I,
don't
think
I
start
all
right
so
like
what
I
was
thinking
about
is
we
could,
whether
it's
built
into
the
cluster
API
manager
as
a
controller
or
its
sidecar
tool
sort
of
thing?
A
So
if
you
create
a
cluster
or
a
machine
and
you
don't
specify
a
specific
bootstrap
config
reference
or
an
infrastructure
reference,
we
could
have
this
thing
automatically
generate
some
defaults
for
you.
So
it
would
generate
the
objects
and
set
the
references
and
you
would
have
some
sort
of
configuration
and
it
can
pick
map
or
an
actual
type
where
you
could
say
well
in
the
absence
of
any
sort
of
specification,
I'm
using
Kappa.
A
So
here's
my
like
making
an
AWS
machine
and
here's
my
region
and
whatnot,
but
that's
like
going
from
zero
to
a
bootstrap,
config
and
I,
think
you're,
maybe
looking
for
there
drop
config,
but
people
didn't
fill
in
certain
fields.
So
let's
give
them
a
good
experience
because
I'm
using
Kat
V
with
the
central
PKS
or
something
right,
yeah.
H
It's
more
tied
specifically
to
that
cube
ad
and
config,
and
things
I'm
a
file
will
sort
any
anything
else.
I
could
add
there
and
to
the
order
cloud
in
it
yeah,
but
yeah
I
like
that
idea
of
a
controller
I,
think
that
makes
sense
because
then
you
know
we
could
anybody
could
provide
that
controller
and
say
here's
how
you
configure
it.
It
will
apply
those
defaults,
I,
think
that
makes
sense.
Andy
I
like
that
idea.
Thank
you.
H
All
right,
well,
I,
might
I'll
make
a
note
on
this
Andy
that
we
talked
about
it
on
this
issue,
and
maybe
we
could
figure
out
a
way
of
a
path
forward
here
and
I
know
we're
talking
about
it
sort
of
like
what
we
could
do,
but
there
there
is
a
need
for
this,
so
whatever
we
can
do
to
help
get
the
ball
rolling
on
yeah.