►
From YouTube: CAPZ Office Hours 04-17-2020
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
B
A
A
Lady,
thank
you,
David
appreciate
it
all
right.
So
now
we're
just
gonna
go
through
the
list,
we'll
do
our
open
discussion
and
it
will
answer
open
questions
and
the
last
maybe
five
to
ten
minutes.
We
can
go
through
a
project
board
and
just
make
sure
it's
up
to
date,
but
really
want
to
spend
time
on
talking
about
new
things
that
are
coming
up
in
the
cab
Z
project.
So,
let's
start
with
Cecile
go
ahead
with
your
patch
release,
I
think
important
sa.
So.
C
I,
don't
see
Steven
here
so
I
was
hoping
he'd
be
here
and
maybe
circle
back
on
this,
but
I
just
wanted
to
give
her
a
heads
up
that
I
think
we
should
cut
the
new
cut
release
he
fixed
bug.
There
wasn't
any
allowed
on
the
load
balancer,
which
is
causing
tanks
Inc
to
not
work,
so
we
fixed
that
it
semester,
but
I
think
you
should
at
least
said
it's
pretty
important.
So
I
can
kept
the
release,
but
just
wanted
to
see
if
anyone
had
anything
that
they
want
to
also
put
in
there.
C
C
C
D
B
C
C
The
periodic
joke
is
failing
the
last
one
failed
because
there
was
still
the
last
thing
that
went
into
master
this
morning
that
was
required
for
it
to
run,
but
I
was
able
to
test
on
the
TR
by
adding
a
pull
request
or
resubmit
job.
That's
completely
optional
and
it's
not
run
by
default
and
the
last
run
of
that
test
yesterday
completely
passed,
except
that
he
had
to
timeout
before
the
cleanup,
so
I
had
to
increase
its
hang
out
afterwards,
but
the
test
case.
The
test
suit,
has
completely
less
really
didn't.
Use.
C
I
look
forward
to
having
more
data
over
time
like
having
to
see
if
it's
consistent,
consistently
passing
or
if
it's
leaky
or
anything
but
yeah.
So
just
wanted
to
give
a
heads
up
on
that.
C
A
E
So
it's
PR
540
drop
that
into
you,
which
that
oh,
it's
a
thanks
anyway,
PR
548
goes
through
and
creates
kind
of
an
opinionated
opinionated
record
structure
where
we
have
under
templates.
We
have
a
flavors
directory
and
the
flavors
directory
any
directory
within
there.
Other
than
bass
will
turn
into
a
flavor.
So
basically
it's
just
a
customized
directory
and
we'll
run
customized
on
it
and
drop
a
flavor
with
the
name
of
the
directory.
Try
to
make
it
pretty
simple.
E
So
if
you
want
to
create
a
new
flavor,
maybe
a
derivation
of
another
flavor
with
just
some
patches
on
it.
It
becomes
really
simple
and
that
PR
has
an
example
of
transforming
external
kubernetes
provider
cloud
provider
into
a
flavor.
So
yeah,
if
you,
if
you
take
a
look
at
it,
love
to
get
some
feedback
on
it
if
it's
good,
bad
and
different,
if
there's
another
way
to
do
it,
so
any
questions.
E
E
No,
let
me
when
you
get
I
need
to
get
I
want
to
make
sure
that
that
it's
end
end
so
I
need
to
you
spin
up
the
vmss
and
make
sure
that
nodes
are
joined
now
I
haven't
updated
it
from
master
in
a
little
bit.
So
let
me
do
that
today.
I
will
mark
it
as
no
longer
whip
and
I'd
start
to
get
reviews
in
place.
I'm
still
going
to
hold
on
it,
though
yeah
all
right,
cool.
E
Super
happy
to
hear
that
thank
you,
Justin
I
will
I
will
possibly
take
you
of
them
and
reflect
that.
Likewise,
if
you're
hacking
me
on
something,
please
don't
have
the
feature.
Job
also
I
think
cecile
I
anybody
else
working
on
it,
we're
totally
open
to
that.
That's
that's!
That's
fun!
That's
goodness!
Hey
awesome!.
E
E
She
done
to
controller
and
I
followed
through
with
the
instructions
over
there
and
they
they
did
work.
It's
just.
It
killed
my
machine.
So
when
fired
up
from
cluster
API,
you
have
the
provider
directory.
Next
to
it,
it
builds
out
all
the
components
need
for
needed
for
cluster
API
if
your
docker
and
then
it
builds
all
of
the
components
for
your
cluster
API
or
for
Katzie.
So
you
end
up
having
like
you
madman
controller.
You
have
like
you
know
all
the
different
controllers
for
cluster
api
plus
your
provider
and
whatever
other
providers
you
want.
E
So
that's
actually
quite
a
bit
for
a
little
laptop
to
take
on.
So
what
I
want
to
end
up
doing
is
through
out
there
a
tilt
file
for
clustered
for
cap
see
individually.
So
what
it
does
is
it
grabs
down
the
version
of
cluster
API
loads
it
into
the
cluster
and
then
just
only
deploys
out
caps
D,
so
it
only
builds
and
deploys
capcity
iteratively
this.
This
reduces
quite
a
bit
of
the
load
on
your
computer
and
it
should
help
when
developing
they
have
a
tighter
Interpol.
E
E
Thanks
any
questions
Oh.
How
do
you
how
to
use
it,
so
all
you
have
to
do
is
run
make
tilt
up.
If
you
have
kind
on
your
machine,
you
should
be
good
to
go
and
it
will
go
and
build
a
coin
kind
cluster
with
a
local
registry,
and
then
it
will
use
tilt
to
deploy
out
too
that
kind
cluster
and
you
should
have
a
working
dev
environment.
If
you
run
into
problems,
please
feel
free
to
ping
me
open,
initially,
cool,
there's
any
questions.
E
A
great
question:
yes:
so
if
you
go
to
the
developer
setup,
you'll
see
two
options
for
setting
up
tilt
and
option.
One
leads
you
down
the
caps,
the
approach
where
you're
developing
a
feature,
that's
only
for
cap
Z
and
that's
a
really
great
reason
to
use
that
just
until
final
Kempsey.
But
if
you're
going
to
create
a
feature
that
spans
both
KP
and
cap
Z,
then
you
probably
want
to
use
the
other
one
but
they're
in
the
docs.
E
A
A
So
like
this
yesterday
and
I
thought
it
was
really
interesting,
so
allowing
existing
kubernetes
clusters
to
join
and
management
cluster,
so
I
wanted
to
talk
a
little
bit
more
about
that
and
kind
of
kind
of
like
what
use
cases.
If
we
were
to
do
this,
what
things
could
we
knock
out
first
and
what
things
would
maybe
be
harder,
because
I
think
this
is
a
really
really
good
story
for
folks
to
like
move
on
the
cab
Z.
A
lot
of
people
already
have
lots
of
Keira
Knightley's
clusters
and
I.
Think
that's
really.
A
A
So
for
that,
like
I
I,
do
think
this
feature
you're
really
useful,
but
I
guess.
My
question
is
like:
with
this
only
be
provider
specific
I
mean
and
then
I
guess.
My
next
question
would
be
like
how
hard
would
it
be
to
to
make
sure
that
your
management
cluster
could
connect
out
to
workload
clusters
that
are
in
other
environments?
Does
that
even
make
sense,
for
example
like?
Would
you
want
a
workload
cluster
to
span
multiple
environment
like
multiple
environments
or
multiple
providers?
Or
would
you
want
one
workload,
cluster
per
provider?
F
B
F
Depends
on
what
you
want
to
do,
we
could
we
could
talk
about
smaller,
hey
cats.
We
could
talk
about
smaller
use
cases
first,
where
it's
a
same
environment,
the
same
environment
yeah
as
opposed
to
are
rather
like,
let's
say,
let's
say,
same
provider,
same
v-net
like
the
the
clusters
are
close
by,
but
I
mean
ultimately
the
ultimately.
The
point
of
the
management
cluster
is
to
be
able
to
connect
disparity
so.
A
I
think
he
said
that
it's
explicitly
mentioned
in
an
on
goes
out
to
manage,
so
one
of
the
non
goals
is
to
manage
non
cluster
API
provision
kubernetes
conforming
cluster
is
so
like
if
you've
already
provisioned
the
cluster
a
cluster
II
API.
Would
it's
an
on
goal
for
cluster
API
to
figure
out
how
to
join
that
back?
It's
only
for
you,
clusters
created.
F
F
A
A
How
much
harder
would
it
be
to
also
join
like
a
GCP
cluster
or
two
management
cluster
was
already
created
like?
Would
it
be
that
hard,
but
I
guess
like
some
of
the
life
cycle,
things
like
we
would
maybe
have
to
pull
from
the
GCP
provider
like
I,
don't
know
how
that
would
work
for
a
management
cluster
I'm,
just
like
thinking
of
them.
D
F
D
F
Yeah
I
mean
it
that
sounds
feasible.
I
know
that
there
are
some
things
that
have
changed
with
pivot
since
I've
last
looked
at
it,
but
essentially
the
initial
idea
is
that
you
lay
down
you
well.
This
is
I
guess
this
is
a
long
time
ago
now,
but
you
lay
down
an
initial
local
cluster
and
then
you
essentially
ship
the
you
ship,
the
speck
of
the
clusters
under
management
out
to
one
of
the
clusters
that
it
just
provisioned
or
one
of
the
nodes
that
are
just
provisioned
right,
so
yeah
I,
mean
I,
think
I.
F
Think
if
we
care
about
this,
the
first
focus
would
be
homogeneous
clusters
as
much
as
possible.
So
same
provider,
same
similar,
specs,
right
and
and
locality
would
look
matter,
I
think
and
then
expanding
out
to
other
use
cases.
But
I
do
think
it's
something
that
we
should.
We
should
look
at
in
the
in
the
upstream
cluster
API
meetings.
C
The
way
I
so
I've
looked
at
cluster
cuddle
move
recently,
that's
the
one
that
moves
the
management
to
another
cluster,
and
so
what
it
does
right
now
is
it
takes
the
so
it's
two
steps:
there's
a
cluster
petal
image
on
the
new
on
the
workload
cluster,
so
that
it
becomes
ultra
API
from
a
culture
and
then
it
moves
all
the
see
our
DS.
But
that
assumes
that
the
CRD,
not
the
series
with
the
resources
that
assumes
that
the
resources
already
exist
somewhere
right
and
you're,
just
moving
them
to
a
new
culture.
C
Copying
them
over,
so
I
think
it'd
be
slightly
different,
because
the
difficulty
of
adopting
a
cluster
that
wasn't
built
with
the
cluster
API
is
that
there
are
no
site
resources,
definitions
out
there
to
define
what
that
cluster
looks
like
so,
you'd
have
to
inspect
the
state
of
the
cluster
and
then
backwards,
like
reverse
engineer,
build
the
CR,
DS
and
so
I
think
that's
the
difficulty
and
the
way
I
see
it.
It
could
probably
work
with
like
a
new
cluster
cuttable
command.
So
it's
not
like
providers.
Specifically,
there.
B
C
Need
to
be
like
provider
aware
environment,
so
you
need
to
say,
like
this
is
an
infrastructure,
Azure
cluster
and
then
you
have
to
really
close
your
cuddle,
duct
and
then
passing
the
credentials
of
that
cluster.
So
we
can
access
it
and
then
sort
of.
Like
reverse
engineer
what
the
cluster
looks
like
into
resource
definitions,
yeah.
A
C
So
there's
two
different
things
which
installs
this
like
the
like:
CR
DS
and
the
controllers
and
makes
it
like
enabled
to
provision
work,
alert
clusters
that
doesn't
do
any
like
actual
provisioning
of
infrastructure.
And
then
there
is
like
move
or
yeah
like
config
cluster
and
apply,
which
actually
creates
like
custom
resources
and
then
that
those
get
translated
into
like
a
sure
or
any
infrastructure
provision.
A
C
C
Thing's
cluster
in
the
world
that
knows
about
faster
API
that
was
built
with
some
other
way,
and
you
want
that
to
become
one
of
your
workbook
clusters,
not
one
of
your
managing
cultures,
yeah
cluster,
it's
different
from
minutes,
because
it's
not
about
making
it
a
management
cluster.
You
could
manage
it
from
another
cluster.
It
doesn't
have
to
be
managed
from
itself
yeah.
A
F
B
A
Yeah,
thank
you.
This
makes
a
lot
of
all
right.
Anything
else.
I
think
this
is
definitely
something
we
could
bring
up
to
copy
and
at
least
get
it
from
an
on
goal
to
a
goal
as
a
first
step
and
then
kind
of
figure
out
like
where
we
looking
now.
If
you
all
can
decide
on,
let's
try
and
figure
out
what
it
means
for
just
within
a
provider.
First,
maybe
get
some
great
thinking
on
it.
So
yeah
any
last
thoughts.
C
Yeah
now
that
Stephen
is
here
just
want
to
bring
your
attention.
The
first
item-
Stephen
they're,
really
that
release
or
the
bug
that
we
found
just
wanted
to
give
you
a
heads
up.
I
will
look
at
doing
that
in
the
next
few
days.
I'm
the
surgeon
yeah.
A
A
A
C
A
A
C
There's
a
PR
open
for
that
one.
It's
right,
blue
and
yeah
on
hold,
because
we
want
I
think
how
do
you
open
that
Karen
he's
here
actually
but
I
think
Steven
was
going
to
take
a
look
at
it.
Mm-Hm
yeah.
A
A
A
F
C
Yeah
this
is
three
days
ago,
so
it's
pretty
recent.
He
just
expressed
his
interest
to
work
on
it
to
me
and
I
just
assigned
them
an
issue.
Someone
else
cool
cool
sounds.