►
From YouTube: 20190501 kubeadm office hours
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
Our
agenda
is
pretty
light.
So,
if
you
have
details,
please
add
them
to
the
above
magenta.
Would
you
have
a
thorny
topic,
which
will
probably
take
a
long
time
for
us
to
discuss,
but
the
first
one
we
have
up
was
carry
over
from
last
time
and
I
don't
see
for
breed
CEO
here
with
regards
to
this
issue,
so
we'll
just
punt
that
one
in
the
dealership,
we
can.
You
can
talk
about
it
next
up,
Lemire,
disable,
static,
pods
and
worker
notes
by
default.
A
Can
I
don't
there
was
the
weird
thorny
issue
with
regards
to
skipping
phases
and
that
like
you're
out
of
band
you're
you're
no
longer,
and
it's
like
this
support,
loop
I,
don't
think
we
should
disable
this,
like
I,
think
we
should
enable
the
use
the
user
story
I.
Think
it's
common
I,
don't
know
whether
or
not
we
want
to
do
it.
This
cycle
I
know
we
we
pontificated
not
doing
this
in
the
past,
we
said
look.
A
This
is
something
we're
not
going
to
do
because
it
gets
us
into
a
weird
scenario
but
I
think
so
long
as
we
sort
of
taint
a
node
in
some
way,
shape
or
form.
That
says:
they've
gone
off
the
beaten
path
and
then
future
upgrades.
You
know:
they're
they're,
not
they're,
not
technically
supported,
we'll
do
our
best
effort,
but,
like
you
know,
you're
on
your
own,
this
phase
out.
B
Think
we
should
do
it
for
asset
and
upgrade
and
I
mean
from
there.
Whatever
users
are
going
to
curate
it
it's,
we
can
have
some,
like
you
say,
a
paint
or
basically
dogs
that
tell
them
that
they
should
be
careful
about
the
interest
Mario
that
is
not
supported.
I
think
we
should
do
it
in
terms
of
the
consistency
of
our
main
commands.
C
A
Don't
see
how
we
cannot
do
both,
though,
do
it?
That's
a
good
question.
I
think
for
the
group
do:
do
we
want
to
have
the
capability
to
have
arbitrary
execution
of
phases,
or
should
we
just
want
to
skip
phases
for
those
two
sub
commands
and
it
makes
sense
to
have
the
arbitrary
execution
of
phases,
but
there's
only
so
many
things
that
we
do
in
enjoy.
B
A
B
A
B
A
B
Cards,
yes,
so
Josh
also
created
a
ticketing,
cube
adium
and
basically
he
explained
the
scenario
where
I
mean
you
can
write.
Static,
pods
static,
pod
manifests
if
a
certain
work
world
is
given
special
privileges,
and
this
can
spawn
a
bunch
of
pots
that
basically
he
he
is
proposing
that
this
is
kind
of
a
security
concern
and
also
Duffy
said.
That
is
probably
a
good
idea
to
disable
it
by
default.
But
then
again,
it's
kind
of.
If
you
give
this
work
for
such
privileges,
I
mean
the
work.
What
can
do
very
much
was.
A
So,
like
you
know
you
have
to
be,
you
have
to
be
admin
level
to
run
Covidien
right
in
order
for
you
to
actually
drop
the
static
manifests
and
if
a
person
overrides
aesthetic
manifests
into
that
location,
that's
totally
up
to
them
and
there's
a
bunch
of
bootstrapping
scenarios
that
can
exist.
That
is
highly
beneficial
to
have
that
static
minimis
because
before
it
actually
joins
the
cluster,
it
might
need
to
do
some
initial
behavior,
right
and
I.
B
A
A
B
B
A
Why
do
we
do
that
and
then
we
can
come
back
to
it
and
I'll
write
up
what
I
just
said
earlier
after
this
meeting?
Okay,
all
right,
so
this
is
the
one
that
will
probably
take
up.
Most
of
the
meeting
is
Ross
on
the
call.
Yes,
he
is
you
know.
Ross
I
and
Lumiere
had
lots
of
comments
on
this,
and
so
did
I
said
II
from
ratio
is
not
here,
but
he
also
had
a
bunch
of
comments
too,
as
well.
I,
don't
know
if
he
actually
put
them
into
the
dog
ran
out
yet
so.
A
There's
other
pieces
inside
of
here
that
I
we
can.
We
can
sort
of
walk
through
this
and
sort
of
outline.
What
partials
we
disagree
with,
but
I
guess.
My
question
is:
for
the
authors:
do
we?
How
do
we
want
to
proceed
here?
Do
what
I,
like
kind
of
walk
through
the
contentious
bits
here
from
top
to
bottom
and
see
what
we
can
do
and
see
what
we
can
agree
on
and
disagree
on.
E
Before
we
get
into
the
specific
side,
I
think
I'm
trying
to
understand
what
the
possible
outcomes
are
here,
because
I
think
some
of
the
things
that
there
were
disagreements
around
or
I
kind
of
wonder
whether
or
not
they're,
even
within
the
scope,
it's
a
closer
lifecycle
or
if
it's
you
know
just
simply
a
windows,
bootstrapping
issue.
Is
it
something
we
should
just
take
out
of
the
cap
and
solve
on
our
own
within
sig
windows
like
wait,
you
know
with
potentially
splitting
the
work
up
and
up
what's
clearly
the
responsibility
of
the
key
Bay
Deanna.
A
That's
what
we
can
use
as
a
reference
point
to
talk
about
some
of
these
issues.
Yes,
sir
yeah,
because,
like
some,
some
of
the
bits
are
our
interesting
like
how
you
initially
bootstrap
was
flannel.
I
kind
of
wish
for
BG
was
here,
because
we
had
lots
of.
We
had
a
very
long
conversation
last
time
on
this
topic,
all
right.
So,
let's
start
walking
through
here.
A
G
E
I
think
one
of
the
reasons
that
were
interested
in
the
script
is
it's
something
that's
transparent,
that
an
admin
can
review,
because
I
wouldn't
expect
them.
I
mean
like
if
someone's
going
to
be
turning
this
into
a
part
of
you
know
an
enterprise
deployment.
I
wouldn't
expect
the
Windows
admin.
That's
receiving
this
to
go,
be
able
to
review
the
goaline
code.
A
H
A
A
B
Only
modification
that
we
need
there
is
a
single
sentence
to
the
top
that
pretty
much
breaks
the
further
reading.
If
they
want
the
users
want
windows,
control
planes,
I
mean
Windows
worker
nodes,
but
most
wrapped
with
a.m.
they
should
go
to
a
different
location.
That's
the
that's.
The
official
talks
for
Windows
support
at
the
moment
where.
I
B
B
A
B
B
H
So
so
so
the
existing
implementation
will
break
right.
So
are
we
saying
that
it's
not
possible
for
us
to
do
a
cube,
ATM
upgrade
where
we
actually
fix
both
the
the
sequence
of
steps
and
execution
of
those
steps
on
Windows
so
that
it
doesn't
break
so
I
mean
the
important
thing
I
want
to
mention
here
is
that
we
don't
want
to
see
the
process
and
the
sequence
of
events
for
Windows
doesn't
have
to
follow
the
exact
process
and
sequence
of
events
on
Linux
that
I
mean
the
different
operating
systems.
They
have
different
requirements.
H
B
H
Think
that
that's
an
option
services
is
the
best
way
to
run
the
process
on
Windows,
to
make
sure
that
it
has
resiliency
and
continuously
stays
up
and
running.
So
I.
Don't
think
that
I
think
that
that's
a
non-starter
but
services
can
be
controlled.
You
can
control
how
they
get
stopped.
You
can
know
how
they
get
upgraded.
You
can
control
how
everything
gets
executed
on
them.
H
E
I
J
J
H
A
H
H
I
H
B
H
It
I
don't
expect
I
mean
you,
you
already
have
a
very
full
workload
right,
so
it's
expectations
another.
They
have
every
single
technical
problem
solved.
At
the
time
of
the
cap,
I
mean
we
have
a
high-level
idea
of
what
we
want
to
implement
and
maybe
we'll
discover
a
couple
of
things
down
the
line
and
we'll
have
to
work
through
them,
and
maybe
you
cause
something
unsupported
for
for
the
Alpha
release.
This
is
an
alpha
release,
so
I
I
don't
want
to
put
some
more
worker
than
you,
lube
Amir,
and
you
feeling
bad
saying.
H
J
Think
the
the
tricky
bit
is
just
like
so
QB
IDM
is
like
designed
with
you
know,
taking
advantage
of
certain
features
to
work
and
I,
don't
want
to
go
down
like
too
far
down
the
road
of
like
wildly
diverging
that
approach
for
Windows.
You
know
like
if
it
just
starts
like
scripting
processes
on
the
host
instead
of
like
using
two
handsets
I,
don't
know
it's
a
big
divergence
of
behavior,
alright.
A
A
I
just
want
to
make
sure
that
we
use
the
same
patterns
that
you
would
use
on,
Windows
that
he
would
use
for
other
deployment
scenarios
and
right
now
we
don't
actually
have
that
knob
to
enable
that
condition
and
I
think
it's
fine
to
putt
for
this
cycle,
because
we
don't
actually
have
the
upgrade
capability,
but
I
think
we'd
absolutely
have
to
have
it
for
the
next
cycle.
So
I
don't
think
it
should
be
an
on
goal.
I
think
should
be
like
a
phased
implementation
approach
or
at
least
say
for
initial
creation,
or
something
like
that.
A
B
A
E
Mean
it's
gonna
be
based
on
whether
or
not
people
ask
for
it
and
I
mean
I,
do
think
that
once
people
start
picking
up
Cooper
days
for
Windows
with
Windows
nodes
and
then
I'm
expecting
that
there's
going
to
be
some
shops
that
are
mostly
Windows
today,
they're
gonna
ask
that
question,
but
we
just
need
to
see
whether
or
not
we're
gonna
get
or
not.
That's
going
to
be.
A
E
A
C
A
We
talked
about
the
script
execution
of
the
script
and
I
think
there.
His
main
issues
here
about
the
the
workflow
that
was
outlined,
I
think
that's
reasonable
committee
I
managed
to
the
start-stop
of
the
Kulin
service.
This
is
the
initial
startup
of
the
Cugat
itself,
and
this
is
where
it
gets
a
little
bit
weird
with
the
details.
B
So
we
already
have
the
wrapping
the
multi-os
wrapping
it's
done
and
was
somewhere
in
the
package
for
the
so
the
package
package
under
in
it
system.
So
if
the
operating
system
that
currently
is
running
on
the
load,
if
we
do
start
service
couplet,
it's
going
to
behave
differently
between
lindo
windows
and
linux,
that's
already
supported.
The
question
is
how
we
propagate
flags.
A
So
I'm
just
reading
here
could
be
near,
is
quite
coupled
to
system
tea.
At
the
moment,
this
file
shows
some
of
the
couplet
flags
for
a
better
Linux
bound.
So
it's
basically
the
Linux
overrides
are
the
the
overrides
for
the
clip
flags.
Today
we
have
multiple
mechons
to
have
different
couplet
configurations
for
joined
for
worker
joins,
so
you
can
specify
multiple
config
maps
for
the
couplet.
This
is
for
the
initial
couplet
deploy.
E
A
So,
like
there's
an
initial
lay
down
from
the
package
manager
yeah
that
exists
right.
So
that's
what
you'd
have
to
manage
for
your
initial
couplet
deploy,
but
then
there's
a
there's.
These
flag
overrides
that
exist
inside
of
that
lay
down
that
say:
go
look
over
here
for
the
configuration
file
move
it,
which
is
the
sort
of
it's
not
full
of
dynamic,
complete
cupola
configuration,
but
it's
the
these
are
all
the
flags
that
you
need
to
run
your
Google
Earth
and
that's
what
committee
M
lays
down
for
you
right
and
now.
B
So
the
concern
about
the
flaxian
geno
is
correct.
We
should
move
away
from
writing
the
file
with
the
flags
as
soon
as
possible.
We
have
a
flag
found
basically
that
qadian
populates
dynamically
and
like
I,
spoke
to
my
family,
basically
saying
that,
like
these
flags
are
going
away,
so
the
the
way
to
configure
the
coverage
should
be
using
the
complete
configuration
yes
and
I
think
that
we
should
do
the
same
on
Windows.
B
So,
basically,
we
need
to
abstract
the
Linux
way
that
is
kind
of
currently
wrong
and
deprecated,
and
we
have
to
do
the
the
windows
way.
My
suggestion
is
going
to
be
to
use
the
the
complete
configuration
instead
of
any
flags
unless,
unless
we
face
some
sort
of
a
blocker
where
we
cannot
write
the
configuration
properly
in
which
distress,
but
there
has
to
be
an
abstraction
there,
because
we
cannot
use
the
current
way
of
basically
managing
the
flags
for
both
operating
systems.
A
A
There's
two
parts
right,
so
what
he's
saying
is
there's
there's
a
config
map
that
exists
here
and
then
what
join
does
is
join,
will
basically
read
that
config
map
and
then
given
some
other
data
in
here
well
then
write
the
file
right
right.
That
file
is
the
override
that
exists
right
on
the
worker,
node,
okay,.
A
I
A
A
E
A
That
could
the
configuration
will
probably
need
to
be
verified
so
part
of
the
init.
What
you
could
do
is
you
could
have
a
not
a
bad
process.
The
out-of-band
process
would
be
like
somebody
needs
to
push
a
Windows
config
map
and
Kuby
TM
joy
needs
to
specify
config
map,
and
that
can
think
that
needs
to
be
your.
Your
your
pre
pushed
populated
config
map
for
Windows,
ok
and
who
pushes
I
config
map
for
Linux.
That's
done
on
an
it
by
default.
A
E
B
Definitely
so
we
on
Windows
when
the
node
is
joining,
we
can
join
it,
but
we
actually,
we
have
to
consider
the
the
corporate
configuration
that
was
created
by
NIT,
but
we
have
to
apply
the
Windows
defaults,
so
we
basically
have
to
mutate
the
the
corporate
configuration
from
the
official
config
map.
Ok,.
E
A
A
A
Possibly
I
mean
the
one
exit
strategy
is
that
part
of
your
init
script
is
to
drop
this
and
push
this
config
map
right,
giving
the
right
permissions
so
that
a
person
can
do
this
specific
override
that
this
data
has
never
used
the
defaults
are
never
used.
You
have
a
specific
config
map,
overwrite
that
the
Reds
to
a
file.
I
Yes,
it
meant
I
only
want
that
they
want
to
highlight.
Is
that
currently
what
we
do?
One
note
that
is
to
use
environment
variables
as
a
probe,
because,
basically,
the
cupola
two
node
links
the
config
map
in
the
cluster
that
is
shared
across
of
node
and
user
and
variable
environment
variable
to
apply
node
specific
parameter,
and
these
are
low
basically
to
share
a
configuration
between
multiple
nodes
and
you.
If
you
update
this
configuration,
then
restart
the
nodes
you
can
apply.
The
same
change
hold
the
node.
I
F
Alright,
so
there
are
a
queue:
okay,
I
think
that
there
are
a
few
facts
that
are
not
going
away
pretty
soon
because
they
are
not
in
the
comprehend
Kovach
of
qubit,
for
example
the
opt-in
for
a
flag,
so
the
basically
the
the
pose
image
overwrite
flag
for
the
qubit
is
not
going
away
because
it's
basically
part
of
the
doc
regime.
So
also
the
fact
that
there
are
part
of
the
Doc
resume
of
cubelet
are
not
present
in
the
component.
Conviction
not
going
away,
but
the
doctor
regime
is
going
away.
A
This
making
these
changes
we'd
have
to
go
through
like
a
legitimate,
decent
review
to
make
sure
that
we're
doing
the
right
thing
for
long
term.
Sustainability
like
this,
this
style
of
work
seems
totally
reasonable.
This
cycle,
like
what
I
want
to
make
sure
we
do,
is
I,
don't
want
to
overburden
everybody
with
one
specific
agenda
like
we
have
a
limited
reviewer
band
with
an
area
of
a
full
release
right.
A
The
the
this
cap
came
in
at
the
end
of
our
planning,
but
we
already
planned
right
with
your
planning
and
we
already
have
a
full
agenda
and
I
want
to
be
mindful
of
what
we
can
accomplish.
Given
certain
reverb
were
close
and
I.
Do
not
I
repeat,
do
not
want
Lumiere
working
14
to
16
hour
days.
We
can
get
the
code
in
place.
We
can
talk
about
it,
but
I
don't
want
people
like
doing
heroics
to
try
and
get
something
in
place.
H
H
You
know
I'll
find
1.15,
but
you
guys
will
be
willing
to
at
least
offer
some
guidance
and
direction
also
working
on
it
and
if
it
makes
it
in
because
what
about
to
review
it
good,
but
if
it
doesn't
make
it
in
because
you
guys
like
that,
if
your
bandwidth
then
we're
just
pushing
it
through
to
the
next
recycle,
but
at
least
were
every
step
of
the
way
so
making
decisions
and
trade-offs,
you
guys
are
at
least
engaged
and
nobody
so
can
catch
it
early.
Exactly.
A
H
Mean
under
those
assumptions
what
I
would
say,
why
not
allow
us
to
mark
this
as
implementable
for
1.15?
If
we
liked
it,
if
your
bandwidth,
then
we
just
won't,
do
it
and
we'll
all
agree
to
that?
But
if
we
are
able
to
get
a
team
and
the
reviews
are
going
smoother
than
expected,
then
we
can
make
it
even
if
not
it
doesn't
yeah.
H
I
mean
if
things
go
smoothly
from
design
choices
and
the
things
that
Kalia
is
is
working
on.
It
is
possible
that
we
might
be
able
to
to
work
through
this,
but
one
thing
you
mentioned
earlier
that
you
know
this
came
in
late
in
planned
that
you
and
I
had
talked
about
this
almost
like
over
a
month
ago,
if
not
more,
and
we
did
say
that
the
woman
was
gonna,
have
some
portion
of
his
time
assigned
to
help
us
get
the
cube
idiom
for
Windows.
H
A
H
Guys
have
a
lot
of
other
commitments.
I
just
wanna
make
sure
allow
us
to
make
forward
progress
every
time
we'll
have
to
make
technical
decisions
to
make
sure
your
team
is
engaged
to
the
extent
that
you
guys
come
to
my
feedback
and
and
and
and
course-correct,
we'll
make
sure
we
adjust
based
on
that
feedback
and
then,
if
at
the
end
of
114,
were
able
to
at
least
deliver.
H
Maybe
a
couple
of
this
experience,
maybe
a
cube,
a
DM,
joy
and
and
a
couple
of
other
things,
then
we'll
still
call
it
alpha
if
it
hops
people
like
the
biggest
problem,
we're
solving
is
that
is
hard
for
people
to
humble
windows,
nose
to
kubernetes-
and
this
is
this-
is
a
choice
of
making
and
we
want
to
follow
the
standard
the
kubernetes
community
has
adopted.
So
even
if
we
don't
deliver
the
100%
of
coolant
to
an
experience,
every
single
step
that
you
deliver
is
better
than
nothing.
A
K
B
Thank
you
I
want
to
make
a
comment
in
terms
of
like
what
we
want
to
support,
incubate
a.m.
long
term
and
basically
I
as
someone
who
wants
to
maintain
this
window.
Support
help
with
the
maintenance
I
think
that
we
should
ideally
have
the
goal
to
have
Q
proxy
as
a
daemon
set
and
comply
with
what
Linux
is
doing
at
the
moment.
B
If
the
some
customers
do
not
want
Q
proxy
as
a
daemon
set,
then
we
are
entering
the
zone
of
out
of
scope
of
cube
ATM,
because
we
don't
want
to
maintain
long
term
what
it,
basically,
the
of
the
way
of
maintaining
Q
proxy
as
a
Windows
service
from
qadian
I
see
this
as
a
temporary
solution,
but
long
term
I.
Don't
think
we
should
do
it.
L
E
H
So
that's
the
reality
of
Windows
right
so
I
mean
it's
not
something.
That's
gonna
come
out
and
absolutely
nothing
that
will
come
out
in
any
within
a
year
or
so
I
mean
no
we're
asking
for
fundamental
changes
to
be
done
to
the
operating
system,
then
in
the
SIP
windows
to
adopt
out
and
then
deliver
that
in
kubernetes.
Not
you
know
would
dealing
with
an
operating
system
that
has
cycles
that
lasts
for
a
pretty
long
time.
So
we
just
wanna
be
realistic
here
about
expectations.
Yeah.
B
A
Yeah,
why
don't
we
say
that
is
correct
at
credit.
There's
the
listing
of
criteria
in
the
cap
for
graduation
to
full
support.
So
if
we
have
it
as
alpha
and
we
leave
it
as
alpha,
you
know
we
can.
We
can
maintain
the
braking
compatibility
to
remember
for
a
period
of
time,
so
just
edit
as
a
crap
graduation
criteria
in
the
bottom
portion
of
the
cap
itself.
I
think
that's
a
reasonable
expectation,
I.
H
Also
think
that
in
certain
cases,
I
understand
the
context
under
what
Obama
is
coming
from,
that
you
know
having
this
uniformity
of
how
you
managing
assets
and
to
proxy
in
this
case,
makes
sense
to
have
to
have
that,
but
the
spirit
of
the
request
and
sometimes
you're
gonna
encounter
difficulties
in
this
case
with
Windows.
It's
just
differences
from
the
different
assumptions
under
the
cover.
So
as
trying
to
apply
these
you
or
maybe
might
not
make
sense,
long-term.
A
H
A
But
we
need
to
maintain
and
support
that
route.
So
it's
it's
a
good
user
experience
for
the
community
right
like
we
need
to
have
like,
dare
I,
say
it,
which
I
don't
ever
want
to
dare
say
again,
because
I
wrote
one
a
long
time
ago
would
be
like
an
actual
installer
right,
and
you
know
I'm,
not
a
huge
fan
of
creatives
dollars
but
like
I,
think
that
that's
the
expectation
for
the
user
experience.
I
A
H
No,
it
doesn't
it
actually
doesn't
it
doesn't.
There's
a
community
or
sig
note
that
operate
a
node
for
Linux
and
receive
Windows
that
operates
a
node
for
Windows
and
you've
came
up
to
an
understanding
how
we
wanna
operate
a
node
and
whether
do
it
diverges
from
would
you
guys
think
with
the
communities
right
so
they're,
sick
windows
and
they're
sick
note
and
our
node
operates
how
we
believe
it
should
operate
windows?
H
And
what
we're
coming
here
to
say
that
there
is
no
castration,
told
us
as
well
I,
don't
like
how
you're
operating
your
node,
because
I
want
to
changed.
That's
really
what
we're
coming
down
team
and
I.
Don't
necessarily
agree
with
that.
If
it's
feasible
for
us
to
do
it,
we
will
do
it,
but
I
don't
want
to
make
that
as
a
requirement
for
graduation
we'll
do
the
right
thing
for
Windows
users
and
Windows
nodes
I!
Don't
want
us
to
do
something
just
because
it
makes
it
easier
for
us.
We
should
do
the
right
thing.
K
H
A
B
Is
me
with
the
strong
opinions,
I
think
that
the
group
as
a
whole
found
the
you
know
proxy
as
a
Windows
service,
diverging
from
the
ideal
of
qadian
as
a
deployment
tool.
It's
not
personally
me
I
was
basically
looking
for
a
way
to
make
this
feasible
as
an
alpha
and
I
started
the
document
originally.
So
it
was
my
idea
to
patch
something
up
to
get
an
alpha
release.
I
think
we
still
can
do
it.
I
also
think
that
we
should
do
some
of
these
three
factors,
regardless
of
because
you
know
we
have
problems
currently
incubating.
B
We
have
to
show
them
and
by
solving
them
we
are
going
to
enable
Windows
support
one
step
at
a
time.
Oh
I
I
cannot
comment
on
the
governess
whether
this
Cape
is
knowledge
it
or
whether,
if
this
is
a
viable,
you
know
correct.
Alpha,
state
or
I
cannot
comment
on
these
topics,
but
I
think
that
we
can
get
something
working.
That's
mine!
That's.
A
H
We
don't
want
to
shy
away
from
that
right,
so
I
go
into
beta.
We
need
to
make
sure
that
we
have
all
the
tests
and
and
processes
in
place
to
validate
that
this
works
very
well
there.
How
the
sausage
is
made
underneath
might
differ
a
little
bit
and
that's
basically
the
flexibility
that
we
wanted.
You
guys
to
afford
us.
I
Container
I
think
that,
in
my
opinion,
is
okay
to
have
a
Cooper
proxy
to
run
as
a
demon
said,
but
it
should
be
clear
that
cubed
means
to
stay
agnostic,
so
if
it
is
run
outside
the
coop
I
mean,
let
me
say,
management,
it
is
a
pre
require.
We
just
check
that
the
proxy
the
proxy
service
exists
and-
and
we
started
like
we
do-
for
Bay
Cuba
based
on
on
Linux,
so
service
is
our
outside
the
scope
of
cubed
mean
and
also
Staller
are
outside
the
scope
mean
I
think
this
is
the
point.
I
Otherwise
we
are
bringing
equipment
mean
into
a
mess
of
distribution.
This
is
a
one
point
and
second
I
have
a
question.
Are
you
planning
to
use
Windows
nodes
windows
also
for
the
control
plane
for
starting
the
contrary?
No
okay,
because
changing
the
width
will
be
not
contemplated
and
I
think
that
this
will
be
a
constraint.
H
H
A
H
I
don't
know
when
the
planning
meeting
happened.
I
know
we
were
engaged
in
with
with
Lou
Berman
for
a
while
now
I
didn't
know
when
the
planning
meeting
happened,
but
the
idea
with,
if
you
you
know
Callias,
probably
could
not
be
a
little
bit
more
involved
in
your
sig.
So
we
look
at
the
heads
up
next
time.
E
H
Clear
one
thing
be
very
clear
about
what
that
alpha
will
include
right,
so
alpha
will
include
XYZ
join
this
and
that
that
way,
it's
like
what
what
Tim
mentioned
that
time
bounding
on
the
scoping
is
there
that
whoo-ha
bus
also
train
bound
what
the
reviews
will
look
like,
and
then
it
will
also
time
bound.
What
would
you
come
believer?
Yeah.
E
Okay,
I
got
a
really
quick
tactical
question.
If
you
have
time
for
it,
and
that's
directly
related
to
this,
I
was
I
noticed
that
many
Q
now
relies
on
cube,
ADM
Amit
I'm
in
it,
but
I
could
not
find
a
way
to
get
it
to
work
with
flannel.
It
would
be
a
good
person
for
me
to
talk
to
about
that,
because
if
we
have
that
working
I
think
that
could
help
us
bootstrap
this.
This
effort
and
experiment
quicker
by
being
able
to
spin
up
a
single
Linux
node
very
quickly
talk.
A
E
A
Okay,
we
are
out
of
time,
so
we're
gonna
have
to
take
it
to
slack
in
the
mailing
list
yeah.
So
they
focus
have
questions,
comments,
complaints,
concerns
fuel
three
to
focus
on
slack
and
we'll
try
to
we'll
try
to
get
this
thing
turned
around
as
quickly
as
we
can
and
try
to
get
something
into
for
115.