►
From YouTube: Cluster API Azure 2019-12-05
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
We're
in
yoga,
okay,
we're
recording
now
so
welcome
everyone.
This
is
the
sync
cluster
lifecycle
cloud
provider.
Azure
meeting
welcome
to
everyone
who's
here
for
the
first
time.
This
meeting
is
under
the
stance.
Yes
could
have
come
back,
so
please
be
kind
to
everyone
and
respect
each
other,
so
I
think
we
can
start
with
introductions
that
gives
PMI
new
faces.
Would
anyone
like
to
jump
in
and
introduce
themselves
and
tell
us
a
bit
about
what
brought
you
here?
Oh.
C
B
Nicola
I'm
working
yet
giants
form,
so
we've
been
in
touch
with
you
lately
Martha
from
my
team,
so
I'm
working
team
celestial,
which
is
working
on
the
transform4
Azure,
so
I
joined,
transfer
company
recently
like
one
month
and
a
half
ago,
so
I
thought
it
would
be
a
good
idea
for
me
to
join
these
meetings
and,
in
the
beginning,
I'll
be
probably
more
in
listening
mode.
So
just
to
hear
what
what's
going
on
here
and
like
meet,
you
folks.
D
D
A
Okay,
moving
on
I
guess
this
is
the
point
where
we
usually
look
at
a
project
board
and
test
good,
even
if
I'm
wrong,
yeah.
E
E
A
note
for
anyone
who
has
access
to
AJ
who
can
see
the
agenda
but
cannot
access
it.
If
you
join
the
cig
cluster
lifecycle,
Google
Group
you'll
get
edit
access
to
it.
Please
for
the
people
who
are
on
the
call,
take
a
moment
and
drop
your
names
as
attendees
on
the
agenda.
I've
dropped
the
agenda
in
the
chat.
A
Okay,
I
actually
added
that
one
is
moving
this
meeting
to
an
earlier
time
to
accommodate
Europe
contributors.
So
I
know
there's
a
few
folks
from
giant
swarm
Nikolai's
here
today,
but
there
are
a
few
others
that
were
interested
in
joining
this
eating
and
contributing
to
the
azure
provider.
So
I
guess:
does
anyone
have
any
objection
to
trying
to
find
a
new
meeting
time
a
bit
earlier
in
the
day
that
would
work
better
for
Europe?
Yes,.
E
But
no
okay-
the
yes
is
the
yes
is
for
the
fact
that
we
recently
did
a
poll
to
see
when
a
good
meeting
time
would
be.
So,
if
we're
going
to
do
it
again,
one
I
think
we
need
to
wait
until
2020
or
if
we
do
now.
We
should
give
a
good
wide
berth
between
now
and
the
holiday
season
before
after
well
well
after
the
holiday
season
to
give
people
an
opportunity
to
to
contribute
I
don't
want
to
I.
Don't
want
us
to
see
new
shuffling
the
meetings
until
we
are
really
sure.
E
A
A
A
E
B
B
A
E
A
Cool
okay,
next
item
release,
v1
altitude,
so
I
think
we're
pretty
much
ready,
there's
been
I,
don't
think.
We've
had
many
mergers
recently
to
do
with
outfit
to
and
there's
even
if
you
are
open
to
start
adding
those
new
and
often
read
site.
So
we
should
at
least
start
cutting
a
brunch
for
the
new
one
open
to
really
I.
Think
yes,.
E
So
I
had
I,
basically
tossed
the
repo
at
a
few
new
people
internally
and
said,
go
through
the
docks
and
see.
If
you
can
do
this
and
the
general
consensus
was
they
can't
right
now,
I
think
the
docks
have
diverged
so
much
from
what
view
and
alpha
one
was
now
there's
confusion
between
getting
started
and
the
development
stuff
David
I
know
you
did
a
docs
update
recently,
but
there's
still
some
stuff
that
that
could
be
tweaked.
E
So
I've
got
a
bunch
of
feedback
and
I'm
going
to
there's
a
development
guide,
that's
within
the
cluster
API
book,
and
that
development
guide
is
also
out
of
date,
so
I'm
going
to
tackle
updating
that
development
guide
and
pointing
out
to
our
repo
for
any
things,
I
think
we
should
wait
until
that
gets
done
before
before
cutting
the
branch
and
I
know.
I've
said
that
before
so
as
soon
as
the
117
release
is
out
the
door,
that's
essentially
the
branch
manager
for
that.
So
that's
essentially
been
taking
up
my
time
and
cube
cons.
F
E
A
Got
it
and
I
was
actually
gonna,
bring
this
up
later
and
later
agenda
items,
but
looking
for
a
way
to
help,
and
can
she
do
and
doesn't
really
know
where
to
get
started?
I
think
we
would
definitely
appreciate
a
new
contribution
to
documentation,
maybe
not
that
one
in
particular,
but
just
in
general,
I,
think
the
legal
documentation
and
I
think
that's
actually
a
really
good
way
to
learn
about
how
things
work
and
when
you're
a
first-time
user.
A
You
actually
have
that
really
special
first
time
length
where
you
see
things
the
way,
anyone
who
comes
to
repo
for
the
first
time
to
keep
things,
and
so
you
can
maybe
bright
ducks
in
a
way
that's
easier
to
understand
for
first
time,
whereas
someone
who's
been
contributing
for
a
well
kind
of
take
things
for
granted.
Sometimes
so,
please
help
you
are
leaving
wait.
A
E
A
E
A
H
It's
coming
around
slow,
but
coming
around
I
credentials
are
hard
apparently,
and
I
can
run
this
on
my
Desktop
and
the
moment
I.
By
the
way
we
have,
we
spent
some
time.
I
knew
resubmit
for
the
for
the
repo
had
you
had
a
few
more
jobs
to
the
test,
infra
repository
once
that's
the
ones
that
was
in
place
and
if
I
run
this
end-to-end
test
its
timing
out
so
and
I
don't
have
right
now
a
way
to
figure
out.
H
Why
stemming
out,
my
guess
is
that
creations
had
not
picked
up
correctly
and
the
first
the
cluster.
My
first
step
is
just
will
try
and
infinitely
until
things
just
bail
out
and
so
I'm
going
to
see
what
they
Italy
ask
I,
still
to
see
the
output
or
the
controller
locks.
That
might
tell
me
that
something
is
all
confirm
that
something
is
wrong.
So
quick.
E
Thing
on
that,
so
based
on
the
the
piece
of
code,
the
snippet
that
you
sent
me,
if
you
add
a
if
you
just
add
some
log
output
to
that
piece
to
let
us
know
whether
we're
using
environment
based
or
whether
it
could
or
could
not
find
the
environment
based
stuff
or
a
file
based
credentials.
I
think
that
would
get
you
further.
Okay
right,
just
just
check!
Yes,.
H
C
E
H
E
H
F
E
F
H
F
E
H
I
think
that's
that's
a
status
oh
and
I.
Think
I
noticed.
Maybe
something
is
something
I'm
doing
wrong
and
if
I
lunch
and
if
I
try
to
create
two
machines
at
the
same
time,
instead
of
just
killing
one
and
then
I'm
waiting
for
the
next
one
things.
Don't
move,
don't
make
progress
so
I'm
dying
to
figure
out.
If
that's
something
I'm
doing
wrong
or
it's
a
problem.
We
say
contort.
H
What's
happening.
Instance:
creation
yeah,
it's
if
I,
if
I
launch
true
it
doesn't
make
sense
because
the
examples
works,
but
if
I
had
launched
two
machines
at
the
same
time
and
nothing
happens,
I
mean
I.
I
I
wait
for
the
first
one
to
come
up
the
first
control
plane
if
I
launch
more
than
one
the
next
step,
and
in
my
case
this
in
my
test
thing
so
make
progress,
don't
work,
but
it
may
be
a
problem
in
my
dog
on
my
heat
on
my
oh,
my
god.
Okay,
like.
E
A
C
I
guess
it
is
really
just
to
see
if
we
can
get
to
a
consensus
about
the
two
different
approaches.
So
one
was
to
have
new
resource
kinds
and
an
object
reference
within
the
VM
spec,
but
then
obviously
that
results
in
50
are
these
and
a
bunch
of
instances
those
we
have
to
create
and
then
on
the
other
option
is
we
use
the
extension.
C
G
Instead
of
just
having
the
object,
reference
kind
and
API
be
API
version
fields,
it
also
have
the
fields
for
the
image
reference
itself,
and
then
we
don't
need
to
have
the
actual
reference
or
link
between
those
two.
So
when,
if
everybody
agrees
that
you
know
we
want
to
strive
towards
that,
you
know
you
X,
then
you
know
Richard
the
issues
getting
that
working.
Just
let
me
know
and
I'll
be
happy
to
help.
Okay.
G
Approaches
so
really
I
think
the
the
downside
to
me
is
having
the
separates
here.
It'd
be
actual
object
reference.
We
I
think
the
the
primary
goal
behind
that
the
issue
is
to
because
we
might
have
different
image
image
strikes
in
the
future.
We
want
to
be
able
to
make
that
dynamic.
One
way
is
to
have
a
separate
CRT
and
to
have
that.
G
You
know
that
reference
there,
but
the
other
way
is
just
have
it
be
a
runtime
raw
extension
and
then
you
know
be
able
to
detect
that
time
to
detect
that
type
in
run
time,
and
then
you
know
act
appropriately,
so
I
think
just
the
in
terms
of
UX
having
the
single
CRTs
slightly
nicer,
you
can
look
at
it.
Look
you
look
at
the
Machine
tuskegee
show
you
one
view
cool.
A
A
So
we
published
a
kind
of
a
prototype
image
to
be
able
to
test
for
115
and
at
first
we
thought
that
we
could
remain
secure
using
the
kubernetes
minor.
Very
so
in
this
case
116,
and
then
the
patched
versions
would
be
reusing
the
same
cute.
The
problem
is
that
you
need
to
be
able
to
specify
the
image
that
goes
with
your
specific
kubernetes
version.
E
We
we
should
not
do
that.
We
should
drop
the
kubernetes
version
altogether.
The
SKU
should
be.
The
version
should
be
the
cluster
API
Azure
version
at
any
one
point.
If
anything,
we
should
key
on
the
cluster
API
version,
this
the
cube,
the
kubernetes
patch
release
at
any
one
time
that
should
be
solved
in
the
matrix
that
we
have
on
the
on
the
within
the
repo
right,
so
versions.
A
A
E
E
J
A
A
A
K
Ahead
II
mean
like
what
happens
if
somebody
deploys
a
115
cluster,
for
example,
and
wants
to
upgrade
to
116
I.
Think
that's
like
how
do
you
differentiate
the
image
you
just
say
it's
the
same
image
see
me
like
that's
the
question
here
right
is
like
I
changed.
My
kubernetes
version
should
I
get
a
different
VM
image.
Probably
how
do
you
differentiate
great
good
question.
K
G
E
A
E
I
think
it
would
be
helpful
to
understand,
have
documents
it's
somewhere.
I
know
that
there
is
a
short
version.
I
got
very
confused
going
through
all
of
the
image
documentation.
It
would
be
helpful
to
understand
exactly
what
we
can
and
can't
do
for
certain
fields
and
then
maybe
put
this
up
as
an
issue.
A
Yeah
I
looked
into
this
quite
a
bit,
so
I
can
summarize
my
findings
in
an
issue,
but
the
conclusion
I
came
to
is
it's
really
hard
to
distinguish
like
to
such
an
azure
image
of
her
great
version,
so
any
field
you
have
to
filter
is
this
Q
and
the
Q
is
just
a
string
of
50
characters,
so
we
can
do
whatever
we
want
with
that.
But
that's
the
only
thing
we
can
really
change.
A
A
Right:
okay,
there's
a
Loomis
agenda
item.
If
anyone
else
has
anything
they
want
to
have,
let
us
know,
but
in
the
meantime,
I
noticed
that
there
are
a
lot
of
people
that
are
asking
how
they
can
contribute,
and
we
don't
have
that
many
issues
they're
good
for
first-time
coach,
others
that
are
open
right
now.
So
one
thing
that
you
can
do
to
help
if
you're
looking
to
contribute
is
try
to
help
us
open.
A
My
good
sister
shoes,
if
you
go
through
like
a
six
start
and
first
time
developer
deployment,
and
you
know
those
things
that
are
missing
or
not
working
as
you
expect
them.
Please,
please,
please
open
an
issue
and
let
us
know
Oh,
like
we've,
been
getting
started
on
development
on
this
repo.
So
the
way
don't
hesitate
to
share
like
you're
thinking,
yeah.
L
E
Yeah,
we've
we've
done
a
few
of
those
at
this
point,
I'm
not
sure
if
we
need
to
do
another
one,
but
I
think
that
probably
once
we
drop
the
docs,
we
can
do
that
drop
the
next
release
because
essentially
we'd
be
doing
a
code
walkthrough
of
the
oldest
or
the
most
recent
release
right,
which
is
which
is
old
at
this
point.
So
so,
let's
maybe
you
want
to
open
it?
I.
E
A
E
Overall
I
wanna,
say
over
all
I
want
to
say
thank
you
to
everyone.
I
have
been
under
branch
manager,
stuff
and
wrapping
up
the
release
for
117,
so
I
appreciate
everyone.
Who's
been
stepping
up,
you'll,
definitely
do
flexing
your
approver
muscle
and
everyone
who's
been
working
on
the
repo
in
general
over
the
last
cycle,
or
so
so.
Thank
you.
L
M
N
E
G
E
C
G
A
G
If
the
closer
you
get
upstream
doesn't
exactly
represent
the
exact
abstraction
that
are
kind
of
needed
for
the
various
forms
of
sort
of
managed
clusters,
and
so
those
are
kind
of
I
think
those
are
the
things
to
think
about.
So
I
was,
you
know,
kind
of
one
of
the
few.
Anybody
has
any
opinions
on
this
or
or
they
thought
about
this
already.
E
C
E
C
E
G
Right
I
think
yeah
I'm,
that's
along
the
same
lines
that
I'm
thinking
is
really.
You
know
when
I
talk
about
support
for
a
minute.
You
know
support.
We
manage
it
that
you
know
we.
We
should
have
the
right
abstraction
in
upstream
cluster
API.
In
order
to
implement
the
providers
to
you
know,
manipulate
a
managed
cluster.
You
know,
I
think
one
of
the
one
of
the
properties
that
at
least
the
you
know,
GAE
and
enhanced
clusters
have
is
the.
G
The
control
plane
is
the
control
and
management
is
delegated
to
the
cloud
provider
and
instead
of
you
know,
instead
of
Listeria
can
sort
of
managing
it
directly
and
so
I've
been
trying
to
think
about
what
that
really
looks
like
in
terms
of
cluster
API
abstraction
but
yeah
I.
Think
that's
interesting
in
terms
find
the
intersection
yeah.
C
E
Can
start
to
take
some
of
these
types
and
understand
what
they
would
look
like
interacting
with
the
cloud
provider
right
so
like
delegating
specific
operations
to
the
cloud
provider
right
the
scaling
or
not,
scaling
and
reporting
that
satisfactory
to
the
cluster
in
some
way,
right,
I
think
figuring
out
how
to
to
do
that
and
and
take
all
of
the
pieces,
and
eventually
you
know
a
you
know
right
now
we're
doing
right.
Now,
it's
a
it's
a
single
little!