►
From YouTube: 20181003 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
A
A
A
The
default
way
that
people
typically
try
to
use
cube
ATM
to
custom
configure
their
cluster
is,
they
will
often
use
print
defaults,
a
problem
with
print
defaults
is
it
it
throws
in
everything
and
if
they
want
to
take
that
defaults
file
and
feed
it
into
an
it,
it
won't
work
because
it
has
extra
data
that
is
not
pertinent
to
in
it
and
then.
Secondly,
if
they
try
to
fit
it
to
join,
it
also
will
work
because
it
has
extra
information,
that's
not
perfect,
enjoying
so
on
conversation
on
slack,
there
was
a
couple
different
options
there.
A
B
A
A
C
A
I
think
we
can.
We
can
rally
on
slack
so
you
deal
with
this,
but
the
problem
we
currently
face
is
that
one
twelve
one
is
due
for
release
on
Friday
I
would
ideally
like
to
get
at
least
a
minimal
fix
in
for
cherry-pick
minimal
fix
in
for
112
one,
which
is
this
Friday.
So
that
way
people
aren't
blocked
in
the
community,
so
they
can
use
the
standard
workflow,
that's
already
documented
and
then
for
the
future.
A
A
A
D
A
A
I
know
that
the
my
plan
was
after
the
112
release
is
to
fix,
fix
that
version
in
the
cherry-pick
version
to
be
stable,
stable,
112
or
yeah
stable,
112
1.12.
That
should
fix
the
problem
temporarily.
Iiii,
don't
like
our
versioning,
semantics
and
I
saw
your
whip.
I
was
waiting
on
your
whip
because
I
saw
it
come
and
fly
by
problem
is
I.
Have
too
many
pairs
were
to
review
across
many
areas
and
if
it's
a
whip-
and
you
said
you
weren't
happy
with
it-
I
just
skipped
it
the
the
problem
with
that
is
like
we.
A
We
have
to
maintain
this
little
perturbation
every
single
time.
It
doesn't
make
a
lot
of
sense
I'd,
rather
for
us
to
auto,
insert
to
version
information
and
to
have
like
a
flow
that
basically
says
if
that
tries
stable
one.
You
know
for
the
latest
version,
if
only
if
stable
one
dot
current
version,
that's
encoded
in
the
code,
it
doesn't
exist.
B
D
Yeah,
so
what
I
did
in
my
PR
was
I.
Think
it's
already
ready
for
review.
What
I
did
it's
so
it
very.
That
was
an
aversion
from
the
end
point.
So
it's
if
the
version
is
very
new.
D
A
But
that's
that
the
problem
with
something
like
that,
though,
is
you-
can
get
yourself
into
a
weird
scenario.
You
should
probably
error
right,
I
think
that's
an
error
condition
just
because
of
our
support
matrix
right.
If
the,
if
the
end
point
you
reach
is
115,
you
shouldn't
be
deploying
112
bits
on
there,
because
it's
unsupported.
We
only
support
a
SKU
of
one.
A
A
So
why
don't
we
take
it
to
the
PR?
I
do
know
that
this
is
an
issue
that
we
need
to
fix
on
the
112
branch,
so
I.
The
only
reason
I
put
stable
one
in
there
is
just
so
that
we
didn't
have
to
block
the
release
folks
for
the
release
to
try
and
merge
something
because
it
was
their
first
time
going
through
the
cycle
and
that's
a
it.
We
need
to
kind
of
get
rid
of
this
problem
to
just
fix
it.
A
A
C
C
Basically,
what
what
we
have
equipment
mean
now
is
that
we
have
at
least
two
level
of
phases.
We,
if
you
think,
to
the
served
phases
then
inside
the
services
there
are
many
different
sub
P,
so
I
I
think
that
this
is
something
that
we
want
to
preserve,
and
this
list
of
of
phases
should
be
part
of
the
help
of
the
command.
C
C
C
In
this
proposal,
what
what
is
there
is
is
that
we
can
use
some
flags
after
inside
a
continent
so
and
if
there
is
a
proposal
where
I
can
execute
more
than
one
phase,
so
I
can,
for
instance,
have
an
habit
could
mean
in
it
only
phases
or
apply
phases
and,
as
you
prefer
and
then
give
a
list
of
visitor
that
I
want
to
do.
This
is
of
a
variation
from
what
we
have
now
in
phases
because
now
in
phases.
Basically,
you
can
invoke
only
one
phases
at
time,
so.
A
I,
don't
disagree
with
the
long-term
objective
of
doing
something
like
this,
but
I.
Don't
also
want
to
tie
it
to
promotion
of
kuba
diem
right,
I,
don't
think
it's
necessary
for
us
to
have
this
explicit
flow
I
would
or
this
implicit
flow.
This
is
totally
implicit,
not
explicit.
So
the
I
think
if
we
were
to
well
document
to
the
phases
alone
as
they
exist
today
and
have
the
ability
for
people
to
independently
execute
the
phases
in
a
well-defined
set
order.
I
think
that
will
be
enough
for
most
people.
C
Let's
title
split'
down
the
problem,
two
parts,
and
the
first
part
is
that
why
I
think
that
it
is
necessary
to
move
phases
to
manage
physics
frantic
from
the
same
command?
Is
there
the
reason
behind
this?
Is
that
only
merging
phases
and
at
the
end,
the
commander?
You
can
clean
up
a
lot
of
code.
Oh
yeah.
A
I
think
I
think
if
we
were
to
set
up
these
there
there
are
ways
to
copy
and
leverage
the
same
sub
commands
across
two
locations,
so
that
way
either
ud
doop
the
code
and
you're
using
the
common
same
sub
command
across
both
and
that's
done
all
over
the
place
and
going
examples.
But
if
we,
if
we
do
that
and
have
explicit
execution,
so
it'd
be
comedia
minute,
certs,
right
or
kirby
demon,
you
know
you
could
do
each
individual
phase
for
in
it
has
a
separate
sub
command.
I
think
that
is
better
for
sure.
A
I
think
that
specific
piece
is
separable
from
the
the
implicit
ordering
the
skip
and
basically
the
implicit
ordering
and
the
explicit
skip,
because
this
this
adds
extra
complexity.
I'd
like
to
do
this
in
two
steps
right:
the
first
step
a.
C
A
Yeah
I
think
it's
acceptable.
Yes,
so.
C
A
A
Have
it
you'd
have
a
different
UX
for
what
you're
currently
doing
now,
but
though
the
one
thing
I'd
like
to
tie
into
all
of
this
is
part
of
our
logging.
That
we
have
for
execution
of
different
phases
is
to
make
sure
that
the
logging
is
very
explicit
about
where
we
are
in
in
what
phase,
and
you
don't
currently
do
that.
So
if
we
were
to
actually
write
prefix,
either
prefix
a
tag
which
would
be
like,
you
know,
a
bracketed.
A
No,
this
is
the
this
is
the
phase
were
in
for
every
single
phase
that
we
execute
on.
That's
super
helpful
to
the
user,
because
there's
been
a
lot
of
requests
for
people
to
they
do
their
own
phasing
right
now,
but
it's
difficult
for
them
to
understand
what
are
the
exact
phases
that
I
execute
and
in
what
order?
Okay.
C
E
C
A
C
A
That's
the
that's
the
bit
that
I
said:
I
get
what
you're
doing
with
this
one
and
that's
the
bit
where
I
think
eventually
adding
that
and
having
a
dag
executed,
mark
execution
model
makes
sense
kind
of
very
similar
to
how
cops
works.
But
what
I'm
saying
is
like
we
don't
need
that
for
GA.
That's
like
an
added!
Let's
do
that
later
on.
A
F
A
C
C
A
D
A
D
A
Every
a
Dacian
of
the
cycle,
every
time
we
enter,
we
make
a
change.
There's
this
seg
fault
that
people
have
in
their
brains
about
why
we're
making
this
change.
So
if
we
do
in
an
incremental
way,
it's
easier
to
consume
like,
for
example,
the
splitting
of
the
config
I
know,
people
are
very
confused
by
that,
and
because
we
didn't
think
about
the
UX
and
a
staged
approach,
the
consumption
of
that
type
of
change
becomes.
D
D
A
D
A
D
We
also
if
we
move
at
the
top
of
a
face
at
the
beginning
of
each
face,
if
we
print
a
walk
message
like
saying
this
is
a
face
of
cube,
a
DM
and
I'm,
not
sure
we
should
print
a
number
there.
But
if
we
print
a
message,
then
the
user
can
look
at
the
ordering
in
the
main,
cube
ad
ad
a
minute
that
comes
all
the
phases.
My
point
is
that
we
probably
should
have
a
message
on
top
of
each
face.
So
when
the
phases
start,
we
see
the
phases
executing
in
order.
D
A
So
that
way,
it's
clear
from
a
user's
perspective
if
they
fail
somewhere
halfway
in
a
phase
that
they
can
copy,
that
single
line
of
text
and
give
it
to
developer
and
developer
will
instantly
know
where
it
failed.
And
why
or
where
failed
at
least
otherwise,
if
you
give
the
whole
log,
you
have
to
trace
back
the
details.
A
A
Other
than
that,
other
I,
don't
think,
there's
anything
else
that
I
thought
were.
This
was
the
only
part
that
I
didn't
was
a
little
fuzzy
on.
A
D
C
D
A
C
D
D
We
have
a
my
idea
here
is
to
have
cube
ATM
in
it
phases
or
cube
ATM
phases,
sub
commands,
because
this
gives
us
the
power
to
sandbox
all
the
phases
undercut
sub-command.
Otherwise,
if
you
want
to
use
cube
a
minute
and
a,
how
are
you
going
to
do
it
like
how
are
going
to
distinguish
a
face
under
in
it?
That's
my
question
because.
D
E
D
A
They
are
just
other
other
flags.
Well,
I
see
what
Lumiere
saying
from
from
an
actual
sub
command.
P
flags
perspective.
It's
weird.
If
it
was
a
sub
command
it,
which
could
be
a
minute
phases,
then
sorts
is
a
separate.
Every
sub
command
is
isolated
where,
where
certs
is
one
of
the
phases
and
then
the
args
pass
directly
to
the
sub
command
four
phases
right,
it'd
be
like
kuba
demon,
it
is
the
main
command
or
comedian
is
the
main
command.
A
Sub-Command
sub
sub
command
sub
sub
sub
command
arguments,
because
if
you
were
to
have
a
flag,
if
you
were
to
insert
flag
overrides
for
everything
single
thing
down
here,
it
would
be
very
confusing
if
you
cou
mediaman
it
foo
some
other
option
right
and
then
you
do
phase
certs,
because
the
ordering
and
then
would
be
confusing
right
you
this.
This
thing
here
would
have
to
be
in
some
weird
block
ordering
versus
being
a
sub
command
ordering.
Do
you
know
what
I'm
saying
Teresa.
A
All
the
other
flag,
yeah
yeah,
because
that
way
the
ordering
is
super
explicit
about
where
the
arguments
where
the
flags
live
right.
So
it
would
be
like
this.
So
that
way,
if
you
did
in
it,
you
could
still
do
foo
in
it
and
you
could
do
phase
foo
phase
and
then
you
can
do
certs
right,
because
if
you
just
have
this
flag
here,
then
it
gets
weird.
D
A
C
A
C
A
C
A
A
B
A
D
D
C
D
A
D
C
D
C
C
A
Super
easy
defects-
that's
that's
just
a
problem
that
we
had
from
the
past.
We
we
fix
this
in
soda
blue.
We
have.
We
have
commands
sub
commands
that
percolate
across
major
commands
and
we
just
have
a
separate
function.
That
says,
add
sub
demand,
foo
and
then
the
food
is
tacked
on
to
that
thing.
So
you
it's
easy
to
just
put
it
in
the
library
and
then
say:
add
this
up
command.
C
A
Can
I
can
absolutely
help
with
the
execution
the
I
want
to
get
the
actual
kept
merged?
Its
all
purpose
in
this
conversation
is
to
like
agree
and
commit
or
disagree
and
still
commit
on
certain
things.
We
want
to
get
done
and
then
and
then
break
up
the
work.
So
that
way,
it's
not
we
don't.
You
have
to
rely
on
any
one
person.
A
A
A
A
C
A
So
we
still
need
to
like
rationalize
that
and
I
need
to
go
back
and
look
at
the
notes
that
I
wrote
down
a
while
ago,
because
we
kind
of
did
things
at
the
end
of
cycle
to
fix
some
bugs
with
the
initial
proposal
that
we
had.
Yes,
so
maybe
next
week
we
can
have
some
time
to
sit
down
and
just
chat
about
the
config
changes
and
try
to
get
that
synced
up.
C
A
You
could
just
great
so
just
so
folks
are
aware
some
of
the
other.
Those
are
the
two
big
p0
items
that
we
have
for
this
cycle
is
config
the
beta
and
phases
and
I
think
phases.
Stuff
is
gonna,
be
for
greet
Co
in
an
Luba
mirror
and
anyone
else
who
wants
to
join
in
I.
Don't
know
if
folks
want
to
help
here
or
pitch
in,
but
you
know
happy
to
help
folks
on
ramp
to
this
stuff.
So
this
for
pto,
Ross.
F
A
D
A
A
A
E
I
might
attended
that
meeting
last
week,
so
I'm
putting
together
with
straw
man
on
what
ideal
packaging
release
looks
like
and
what's
gonna
date
for
Friday,
but
will
study
cluster
API
awx
to
do
yeah.
So
I'll
have
a
straw,
man
document,
or
he
might
be
that
detailed
at
some
point
on
Friday
to
look
at
and
that.
A
F
A
About
the
South
Hosting
we
did
so
we
didn't
want
to
remove
it,
remove
it.
We
wanted
to
make
sure
that
none
of
the
logic
in
an
it
or
join
has
any
dependency
on
it
and
it
doesn't
I
think
anymore,
but
we
still
wanted
to
have
a
separate
sub
command
like
a
cuvette,
DM
pivot,
or
something
like
that.
But
once
you
do
this
to
be
like
a
warning,
this
is
a
destructive
change,
you're
getting
off
of
the
beaten
path.
You
know.
Buyer
beware!
You're!
On
your
own
now
so.
A
C
Starting
with
mark
to
think
about
to
out
to
clean
up
the
code
and
then
basically,
we
have
to
this
to
choose
if
to
remove
this
or
not
self
hosting
totally,
and
then
eventually,
when
we
have
more
clear
idea,
we
will
restore
the
code
or
we
are
going
to
make.
Therefore
now
to
move
the
pivoting
code
somewhere
else.
A
Don't
have
strong,
the
community
has
pushed
back
against
me
and
I've
kind
of
I've
I,
don't
have
strong
opinions.
They
would
like
it
to
still
be
a
cuvette
diem,
some
command
right.
So
that's
a
totally
separate
set
command.
That
was
what
one
of
the
requests
that
we
got
feedback
from
from
other
people
just
to
have
like
a
separate.
D
C
A
Actually,
don't
really
want
it
there,
but
I
got
pushed
back,
so
the
the
pushback
from
the
community
and
I
want
to
make
sure
we
listen
to
them
was
that
it's
still
useful
to
a
lot
of
consumers
and
they
would
still
use
it
so
I
think
having
that
option
available
to
them
is
is
probably
still
important.
We
don't
want
to
disenfranchise
folks
and
make
decisions
in
absentia.
So
if
that's
what
they
want,
then
I
think
it's
totally
fine
I!
Think
that's
a
good!
C
C
C
C
C
A
C
A
A
C
A
A
C
A
A
There
are
reasons
why
we
only
use
localhost
for
the
API
server
connection
that
went
beyond
just
connectivity,
so
we
should
probably
out
were
we
have
a
I?
Have
a
hard
stop
I
have
to
exit
this
call?
Why
don't
we
chat
and
and
slack
and
point
to
the
issue-
point
Jason
and
I
to
the
issue
and
chunk
and
we
can
chat
there.
I,
don't
I!
Think
it's
totally
fine,
if
you
do
asserts
and
for
those
two
addresses
localhost
in
API
server.