►
From YouTube: 20190130 kubeadm officehours
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
Behavior,
like
in
case,
if
config
file
is
provided
in
the
command
line,
so
that
some
options
are
taken
from
the
config
and
some
are
not,
and
it's
kind
of
cool
bottom
just
silently
ignores
them,
and
that
there
are
number
of
issues
I
specified
actually
three,
but
and
what
I
would
like
to
discuss
is
how
to
change
this
and
in
some
of
the
issues.
Actually
in
one
issue,
Luba
Mir
suggested
three
options
is.
B
B
Third,
one
is
always
override,
but
also
show
a
warning
like
cubed
map,
lying
security,
defaults
and
I
would
add.
One
more
option
is
to
actually
exit
with
an
error
when
user
specifies
something
that
is
dangerous
and
cube
Adam
can
you
apply
it
and
but
give
user
a
possibility
to
force
those
parameters?
So
if,
if
they
want
to
shoot
themselves
into
the
lake,
we
shouldn't
actually.
C
B
Well,
I
actually
don't
know.
I
I
know
that,
like
it's
well,
there
are
like
issues
with
this
and
I
actually
don't
understand
which
options
actually
are
taken
from
the
config
and
which
are
not
so
if,
if
somebody
can
explain
it
and
explain,
what's
the
what's
the
idea,
because,
like
so
far
I
I
see
that's
it's
like
well,
I
I,
don't
understand
what
what
is
going
on.
B
A
D
A
Yeah,
a
warning
would
be
fine
like
it
and
it
does
that
today
right,
if
you
do
ignore
pre-flight
errors
for
given
value,
you
should
I,
don't
recall
already
now,
I
have
to
double
check
it.
You
did
it
used
to
show
that
you
were
over
writing,
so
this
one
seems
like
it's
a
non-issue
other
than
we
should,
if
we're
not
showing
something,
we
should
show
something
I.
A
D
A
Recall
this
way,
that's
the
reason
for
my
harsh
answer
down
here:
it's
ok
to
do
that
for
Kubb
ATM
to
have
this
parameter
override.
It's
not
okay,
to
pass
that
to
the
unit
file
like
if
you
were
to
pass
that
setting
to
the
unit
file,
you
can't
it's
not
a
cluster
anymore,
because
splitting
doesn't
mean
anything
and
the
cupola
itself
has
no
guarantees
anymore,
so
that
particular
parameter
wouldn't
make
any
sense
to
do
that.
A
A
D
Guess
it's
a
matter
of
when
we
when
we
are
parsing
the
component
configuration
for
the
couplet.
We
always
overwrite
fail
swap
on
true,
no
matter
what
the
user
is
doing
and
I
guess
it
is
proposing
here
or
I.
Guess
it's
my
proposal
to
show
a
warning
that
hey
if
you
want
to
sell
fail,
if
you
want
to
set
fail
swap
on
it's,
not
gonna
work,
we
override
this
value
for
you,
I.
A
Think
wow
in
Canadian
to
deploy
on
a
test
environment
because
I,
but
beyond
being
honest,
this
is
actually
how
I
test
I
actually
ignore,
swap
even
though
I
have
and
I
disable
it
in
the
unit
file
but
I
think
from
a
user
perspective
because
I'm
running
it
on
a
desktop
for
a
lot
of
times
in
my
test
environment,
but
the
allowing
this
for
a
user
is
dangerous
because
the
cluster
isn't
even
conformant
anymore
right,
the
so
I
get
I
get
I
get
this
flow.
I
think,
having
this
just
have
a
warning
is
a
legitimate
thing.
F
B
Do
you
know
all
the
options
that
that
we
have
a
right
so
should
we
do
it?
It's
it's
just
an
example
issues.
So
the
issue
is
that
it's
the
main
issue
as
I
understand
it
that
it's
confusing,
because
user
puts
something
into
the
configuration
file
and
expects
that
the
those
options
will
be
applied,
but
they
are
not.
So
it's
not.
A
That's
a
better
question:
I
do
not
think
I
think
we
would
have
to
do
a
help
scenario
for
every
single
command
line
and
walk
through
it
to
make
sure
it
makes
sense.
But
this
is
one
option:
let's
take
a
look
at
what
a
girl.
What
are
the
other
issues?
You
reference?
This
one
was
1
2,
5
4.
This
is
1
2,
5,
1,.
D
A
D
A
Did
you
perhaps
a
better
venue
for
this
particular
one,
because
you're
not
going
to
get
a
consistent
answer
for
some
of
these
things
would
be
to
set
up
a
separate
meeting,
or
we
can
actually
walk
through
in
detail
some
of
the
configuration
options
and
and
write
down
inside
of
the
help
menu
for
different
things.
What
the
behavior
should
be
our
digital
plan
of.
Did
you
plan
on
working
in
it
this
cycle,
I?
Don't.
B
Know
if
it's,
if
it's
needed
but
I,
was
confused
so
and
I
would
like
to
see
it
like
some
kind
of
an
explanation
and
actually
the
list
of
options.
If,
if
they're
only
true,
that's
not
a
problem
at
all,
so
we
will
fix
or
not
fix
it
or
just
add
warnings.
But
if
there
are
many
III,
don't
actually
know
how
to
find
them.
D
So
yes,
ii,
rfl,
is
posting
a
link
about
these
exactly
yeah.
These
are
the
security
options.
Please
check
the
link
in
chat.
These
are
the
security
option
options
we
override
no
matter
what
options
the
users
already
provided
in
the
component,
config
that
he
feeds
to
in
it.
Okay,
and
we
we
don't
show
a
warning
that
we
override
this.
These
fields.
B
G
A
D
Think
that
we
shouldn't
document
them,
because
we
have
to
maintain
this
extra
document
that
changes
all
the
time
we
might
decide
to
add
more
flex
to
the
list
of
it.
Instead,
if
we
have
a
mechanic
too
warm
about
the
overrides
of
these
specific
fields,
the
user
is
going
to
always
see
this
in
the
command
prompt
in
the
terminal
and
know
that.
F
A
In
the
command
line
overrides
inside
of
the
the
optional
arguments
that
you
specify
for
some
of
these,
some
of
the
other
over
eyes
and
configuration
would
be
different
so
like
the
command
line
here
would
be
specific.
Maybe
the
document
when
you
do
the
help
menu
option
for
proof
layers,
they
would
give
you
the
options
of
what
exists
inside
of
the
code
right.
So
I
wouldn't
do
it
as
a
separate
dot.
Maybe
in
code
documentation
is
what
I
was
thinking.
D
A
A
So
so
maybe
a
good
passwords
here
being
like
one
create
a
little
doc.
This
is
these
are
the
different
options,
and
here
are
the
different
behavior
routes
that
we
see
I
know
you
did
that
partially
doctored
like
beat
try
to
gain
a
little
more
exhaustive
and
then
outlined
like
here's,
the
behavior
that
we
would
want
in
the
ideal
case,
for
the
different
enumerated
options
right.
A
D
A
Yeah
yeah
I
think
it's
just
a
starting
point
and
then
we
can
converse
next
week
about
the
behavior,
because
right
now
you
have
two
sets
and
you
have
at
least
in
sense,
if
not
more
and
I
can't
give
you
a
solid
answer
without
knowing
the
state
space
of
the
different
ones,
because
some
of
the
options,
the
behavior
will
be
different
than
what
we
expect
like
you
can't
have
a
same
policy
across
all.
Yes,.
D
D
D
It's
also
can
be
the
control
plane
endpoint,
but
we
don't
handle
this
as
a
flag.
It's
just
an
unnamed
argument
and
we
are
entering
the
space
of
Cobra
magic
here,
but
we
basically
this
forced
us
to
do
some
gymnastics
in
the
phase
runner
that
I
don't
really
like,
and
also
you
can
look
at
my
link.
That
is
called
here's
how
it's
handled.
D
G
Can
I
give
my
two
cents
on
this,
so
I
think
that
many
issue
are
conflicting
here.
The
the
piece
of
code
to
premier
is
showing
was
there
since
time
and
he
was
eater
who
said
when
we
move
at
the
from
one
booster
token,
from
many
boost
from
any
possible
discovery
in
to
one
only.
It
was
part
of
our
Rasta
ginger
last
cycle,
so
I
will
put
these
out
of
the
way
and
it
is
I
egg
that
will
go
away
when
we
want
better
one
go
away,
so
it
is
a
temporary.
G
The
second
problem
regards
to
Cobra
not
to
the
phase.
Remnar
is
not
specific
to
this
rather,
but
is
regard
to
Cobra,
and
it
happens
when
you
have
a
command
that
use
arcs
and
is
sub-command
in
the
problem
exists
when
your
the
domain
of
values
or
your
arcs
overlap
with
the
name
of
your
sub
command.
In
this
case,
in
the
case
of
join,
we
are
unfortunate
because
the
domain
of
valuate
of
the
arcs,
which
is
a
hostname,
can
overlap
with
the
name
of
sub
common.
G
G
H
From
my
point
of
view,
I
see
I
see
wider
problems
with
the
current
design,
but
I
mean
maybe
that's
something
that
I'm.
Maybe
it's
something
that
is
artificial.
The
examples
that
I'm
going
to
say,
but
if
you
have,
for
example,
one
face
the
top
level
phase
that
takes
two
arguments,
two
unnamed
arguments,
and
then
you
have
two
sub
phases
of
this
one,
and
the
first
of
phase
is
using
the
first
argument
and
the
second
sub
face
is
using
the
second
one.
H
Then
the
open
question
is:
how
do
we
pass
these
arguments
to
these
sub
phases
that
we
need
to
pass
both
arguments
to
both
surfaces,
even
if,
even
though,
if
only
one
one
of
them
is
going
to
use,
one
of
those
arguments
like
the
first
sub
comma
so
face
is
going
to
use
the
first
argument
and
the
second
surface
is
only
going
to
use
the
second
one
or
should
we
pass.
This
is
the
open
question
that
I
have
it's
not
it's
not
really
that
tight
to
kibriya,
but
how
we
are
going
to
pass
this
information
along.
A
Back
here,
first
thing
like
what
is
the
exact
use
case
of
a
joined
phase,
where
the
condition
the
condition
is
that
you're
hitting
because
we
went
through
this
in
the
unit
phases,
where,
like
we,
had
to
have
a
hierarchical
model
for
how
the
parameters
flowed
through.
So
what
exactly
is
is
the
one
that's
having
the
issue,
you
know.
What's
the
perimeter
and
incantation
that
you
have
a
problem
or
that
folks
are
seeing
as
a
problem.
G
G
A
Yes,
so
the
comment:
the
comic
pattern
we
had
before
was
that
if
you
had
some,
if
we
had
in
a
parameter
that
was
common
across
multiple
phases
is
that
you
would
so
be
joined
phases,
and
then
you
have,
you
have
to
be
enjoying
I,
don't
remember
if
it
was
here
or
here,
but
you
have
one
override
and
then
we
apply
to
all
phases.
You
couldn't
you
couldn't
specialized
per
phase.
You
can
only
do
custom
because
we
had
this
long
conversation
forbidden
air
bases.
G
And
the
current
state
is
that
all
day
the
the
flags
applies
only
to
the
top-level
command
and
to
the
lift
command
it
doesn't
apply
to
the
intermediate
command.
So
if
you
have
phase
control
clean,
you
don't
have
phases
in
the
control
plane,
but
you
have
to
use
control
plane
all
so
all
Italy.
If
commander
are
operative,
let
me
say
you
can
do
action.
All
the
intermediate
command
are
only
container.
A
G
G
That
provided
their
common,
the
they
provide.
Okay,
we
agreed
that
we
would
like
to
specialize
faces
in
order
to
get
the
faces,
except
only
what
is
irrelevant
for
them.
Yes,
okay
busy
is
already
implemented
for
flags,
but
not
yet
implemented
for
arcs,
but
it
is
not
a
problem
we
can
do
later
is
not
blocking
now.
F
G
G
G
Okay,
this
is
a
corner
case.
The
bed,
the
the
bad
side
of
these-
is
that
before
we
had
something
that
somehow
help
at
the
user
if
they
missed
iPad,
if
you
write
Phase
II
with
two
II
before
common
and
give
you
some
help
now,
because
the
truth,
the
value
domain,
with
with
the
name
of
phases
and
and
the
earth
name
overlap,
we
are
losing
these
user
user
help.
A
A
There's
a
couple
options:
instead
of
trying
to
like
circle
around
the
modification
or
join
like
we
can
when
in
doubt
you
know,
we
could
have
another
layer
of
indirection
where
we
could
say.
Current
join
is
entering
into
a
deprecation
cycle
and
then
we
call
it
underscore
Julie.
We
call
it
something
else
that
that
we
promote
over
time
and
we
go
through
the
deprecation
right.
A
H
H
If
you
pass
three
arguments,
it's
going
to
work
because
and
you're
going
to
just
ignore
the
second
and
the
third
you're
going
to
use
the
first
one
this
until
that
can
be
also
fixed
on
didn't
utilize
her,
but
I
was
wondering
if
there
was
some
way
and
this
can
be
for
the
future.
So
one
sub
phase
is
able
to
say
I,
expect
to
maximum
arguments
or
I
expect
one
exact
argument,
something
like
this.
So
we
have
this
your
way
to
set
to
tell
the
user
how
many
arguments
they
need,
or
this
sub
face
exactly
needs.
H
A
So
why
don't
we
do
this?
Why
do
we
open
an
issue
with
the
different
phases?
Where
could
we
be
enjoying
in
and
the
current
problem
the
current
behavior?
This
is
exactly:
did
we
promote
a
new
phase
and
then
we
wrote
down
all
the
different
command-line
arguments
as
well
as
the
input
leg
overrides,
and
then
it
should
percolate.
We
should
make
more
sense,
because
right
now
we're
talking
so
abstractly
I,
don't
remember
what
all
the
details
of
the
phases
are
for
join,
but
that
way
we'll
have
an
enumeration
of.
A
G
I
G
D
A
Okay,
I
I.
Don't
because
there's
a
couple
ways
you
can
hash
this
problem,
it's
a
question
of,
like
you
know,
figuring
out
what
just
like,
with
the
other
problem,
figure
maneuver
at
the
state
space
and
numerate
the
options
and
then
we'll
pick
the
from
from
the
least
bad
of
the
options.
That's
usually
the
way
we
go
right.
A
D
D
From
me,
was
it
oh
yeah
it
yeah,
okay,
so
deems
created
appear
to
bump.
Dr.
incubate
am
too
18.0
a
zero
nine
and
we
merge
that
we
are,
and
there
there
was
a
nun
appear
that
fix
fixes
problems
in
the
original
pair
begins.
So
we
are
now
kind
of
arbitrarily
bumping
docker.
We
don't
have
an
twenties,
for
that
I
mean
we
cannot
it
we
can
other
intent,
distinct
abilities
anywhere
currently
been
is
refusing
to
bump
kind.
To
18.0
line
is
claiming
that
this
is
going
to
break
other
comedian,
I'm,
saying
that
it's
not
gonna
break.
D
A
So
we
could
still,
we
could
still
bump
the
version
without
ci
signal,
but
not
make
it
the
default
check
right.
We've
done
that
before.
Actually
because
people
have
lot
have
requested
it
where
the
default
CI
signal
is
still
running
at
1806
or
an
older
version.
In
fact,
we
do
this
for
a
very
long
time.
A
A
F
D
So,
basically,
now
we
these
peers,
that
we
are
doing
they're
simply
adding
support
for
this.
It's
not
the
minimum
supported
docker
version.
We
are
basically
adding
the
version
into
the
unit
test
we
perform
like
I
mean
check,
are
talking
about.
The
unit
is
the
other
day,
but
these
particular
ones
the
system
wants.
So
if
you
have
docker
the
new
docker
now,
it's
not
gonna
fail
for
you,
because
we
have
this
new
version
into
the
list
of
supported
version,
but
we
can
still.
F
A
We
can
always
do
more
sane
things.
We
can
always
refactor
the
tests
like.
We
don't
need
to
maintain
craziness
if
we,
if
we
see
debt
like
we're,
not
underneath
in
the
drumbeat
of
having
to
get
to
G
anymore.
So
if
we
see
technical
debt,
you
know
we
can
absolutely
just
clean
it
up
right.
I
do
think.
Does
anyone
know
what
the
lifecycle
of
1806
currently
is
like
how
long
we
have.
D
A
E
D
A
G
So
today,
I
started
to
write
the
tracking
issue
where
I
put
together
and
if
you
want
to
click
to
the
link
where
I
put
together
the
activity
that
we
have
to
do
for
the
certificate
copy
workflow
and
what
was
already
done
and
also
a
set
of
other
other
improving
possible
improvement.
Some
of
them
are
already
done
for
the
joy
control,
plane
and
so
on.
G
So
there
is,
we
are
in
in
a
good
shape,
Yago
ready
volunteered
for
starting
with
the
certificate
couple
of
row,
but
there
are
room
for
for
many
other
improvements,
so
it
is
a
PSA
I,
didn't
added
to
too
many
detail
in
this
turkey
issue,
because
to
keep
it
simple.
Just
pinged
me
and
I
will
point
to
point
you
to
all
the
details
and.
G
There
is
something
that
we
should
the
site.
I
think
we
should
agree
upon
and
at
the
beginning
before,
starting
which
are
the
change
to
the
configure
for
the
encryption
key
and
basically,
the
contest
here
is
that
we
want
to
allow
the
user
to
use
to
to
set
the
encryption
from
key
in
the
config
for
the
image
workflow.
G
So
the
encryption
key
doesn't
get
logged
in
in
the
history
or
whatever,
and
same
applies
for
day
for
the
workflow,
where
we
want
the
user
to
use
the
the
joint
config
in
order
to
specify
the
encryption
key
for
downloading
certs.
So
we
need
to
change
the
configure
and
I
think
that
it's
important
that
that
we
agree
where
this
changes
should
go
before
acting
in
order
to
not
add
too
many
recycles
on
the
on
the
PR.
G
G
G
The
user
has
to
option
for
for
the
for
the
encryption
key.
He
can
let
to
Padmini
generated
encryption
key
and
then
could
mean
will
print
the
generated
encryption
key
in
the
init
output.
But
this
is
visible
and
this
will
be
logged
or
as
alternative.
The
user
can
generate
an
encryption
key
by
its
own
with
another
command
that
we
have
to
define
then
pass
the
predefined
encryption
key
to
the
we
need
using
a
config
file.
A
G
I
G
A
D
A
A
A
D
And
where
we
are
actually
already
doing
it,
because
we
print
warnings
for
that
and
also
Jordan
participated
in
the
discussion
that
I'm
talking
about
where
somebody
added
a
field
user
in
the
API
machinery
general
complaint
and
ask
the
Jordan
if
they
should
bump
v1
to
something
else
and
journalist,
said
Dolan
said
that
they
technically
should
do
it.
They
should
bump.
A
D
G
G
H
A
A
D
A
No,
it
would
be
stored
on
Cluster
the.
What
Fabrizio
is
saying
is
that
in
it,
config
is
not
stored
in
cluster
there's
an
it
conveyed
in
cluster
config.
What
he's
saying
is
the
localized
init
config
would
be
passed
in
and
it
would
have
that
localized
perimeter.
It
would
never
be
visible
near
command
line.
It
would
only
be
visible
to
administrators
who
can
do
in
it
and
do.
J
J
J
G
A
G
A
Yes,
so
if
we,
if
we
need
to
punt
on
that,
I,
don't
think
it's
the
end
of
the
world
for
storing
a
user-generated
token
it
would
still
be
able
to
get
by,
but
we
could
pass
her
in
a
cycle
to
you.
Why
don't
we?
Why
don't
take
to
that
particular
issue?
Asynchronous,
let's
talk
with
Jordan
about
API
promotional
field,
Edition,
our
pioneering.
A
D
D
We
temporarily
disabled,
that
to
be
able
to
perform
a
gymnastic
in
taste,
test,
infra
related
to
run
after
success.
That's
basically
a
field
in
their
config
for
pro
and
I'm.
Yet
to
see
trail
didn't
have
the
time
to
verify.
This
test
continues
to
run.
I
just
wanted
to.
Let
you
know
that
we
we
are
fiddling
with
the
submit
medium
and
my
idea
there
is
that
that
we
should
probably
leave
it
completely.
A
moved
country-
yes,
I,
will
agree.
A
D
Because
they
forward
different
process
with
a
basal
built
they
acquire
like
see,
I
shake
cross
artifacts.
They
don't
depend
on
the
PR
that
just
built,
but
the
artifacts
were
built
from
a
PR
in
a
piece
of
meat,
so
yeah
it's
much
easier
there.
We
are
I
also
tried
to
explain
to
the
sick
testing
folks
that
we
watch
the
periodic
like
dominantly.
D
A
A
vast
majority
of
us
I'm
fine
with
kindly
whatever
it
is
so
as
long
as
it
catches
the
could
be
the
in
bugs
before
they
get
in
like
last
cycle.
There
was
a
change
that
we
had
to
revert
or
was
it
last
cycle
is
I.
Remember
because
it
was
not
on
the
PR
blocking
jobs.
So
I
thanks
for
the
update
I'm
with
you,
I
agree
with
you,
I,
don't
know
what
else
is
there
anything
else?
You
want
me
to
follow
up
on.
Let's
procrastinate,
no.