►
From YouTube: 20190807 scl 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
Ok
yeah,
so
do
we
have
any
new
participants
on
the
list?
Let
me
check
I,
don't
think
we
have
anybody
new,
all
right.
So
the
next
item
is
PSA.
I
did
a
big
big
grooming
of
the
Canadian
repository
and
I
call
something
like
30
40
issues.
Some
of
them
were
either
stale
or
I
had
a
pretty
good
explanation.
You
know
pretty
good
reason
why
we
should
close
them
and
sorry
for
the
spam,
but
we
are
now
down
to
70
tickets
and
we
used
to
have
something
like
110.
A
A
We
needed
a
top-level
approver,
because
his
PR
is
touching
the
core
modules
file,
because
Kip
Caro
is
now
outside
of
three
and
they
used
to
vendor
customize.
But
now
Kip
Carroll
is
going
out.
So
we
have
to
introduce
customize
again
in
kubernetes
because
of
qadian,
and
this
requires
us
to
update,
go
module
file
and
we
had
to
assign
a
top
level
blower
we
got
Jordan
and
Jordan
saw
some
potential
problems.
Perhaps
it's
already
responded,
but
Jordan
requested
further
review
from
Jeff
Reagan.
A
A
B
Yeah
her
requests
from
the
main
meeting.
The
talk
has
been
submitted
for
ku
vidiian
for
Iago
and
for
bricio
to
do
the
cube
idiom
talk
for
kuku,
North
America,
which
I
think
it's
good.
It's
good
to
have
new
at
all
and
rotate
people
around
and
get
different
people
speaking
just
a
reminder
that
there's
typically
three
venues
a
year.
B
So
you
know,
we've
recently
had
sparse
attendance
on
coop
con
China
with
from
this
thing,
I
think
it'd
be
nice
to,
as
we
think
Ford's
for
next
year
to
get
maybe
some
other
folks
involved
in
the
coop
contra,
no
tux
or
Asia.
Now,
I
guess,
I,
don't
know
if
it's
gonna
be
strictly
China,
it'll
be
Asia.
Now.
A
C
A
C
A
Yeah,
that
was
my
biggest
question
about
the,
because
we
have
a
lot
of
leaf
branches.
I
wanted
to
attempt
to
simplify,
but
then
I.
You
responded
that
my
observation
is
not
correct
about
how
the
dough
stack
logic
is
going
to
work
so
I'm
overall
I
think
I'm
fine
with
this.
So
you
know
I
added
a
proof.
Somebody
else
can
have
a
look
for
LG,
GM
I
think
this
PR
is
pretty
good.
I
only
had
to
I
think
updates
the
variable
name
that
they
requested.
That's.
C
D
A
Yes,
this
is
the
networking
test.
It
only
has
single-dose
type
test.
I
requested
this
sorry,
so
not
spotting
them
earlier
yeah.
This
is
the
new
test.
I
think
is
pretty
good
at
this
point,
let
me
see
what
lame-o
edit
I
saw.
Lg
came
up
roof
at
this
point,
I
mean
anybody
can
still
have
a
look
and
we
can
amend
if
needed
later,
but
yeah.
This
is
good.
A
C
I
would
like
to
keep
working
on
it.
I
think
there's
a
so
there's
a
phase
to
do
will
start
coming
in
and
that's
a
major
PR
going
into
kubernetes
there's
a
associated
set
of
changes
that
need
to
be
made
into
cube,
ADM
and
I'm.
More
than
willing
to
just
keep
added
there'll
be
changes
on
top
of
pod
cider.
For
the
most
part,
it's
just
supporting
a
list
of
siders
yeah.
A
A
D
A
So
we
had
a
very
fruitful
discussion
at
a
point
in
time
with
you
and
rusty
that
we
like
spoke
briefly
about
maybe
dis
validators.
Some
of
them
are
completely
not
needed.
Okay,
baby
I'm,
but
probably
the
first
step
we
can
take
his
indeed,
you
know
have
the
the
separate
rapport
and
then
we
can
decide
what
to
do.
D
A
A
A
Let
me
quickly
open
the
show
one
of
the
piers
by
Fabricio
revealed
that
we
didn't
clean
correctly
one
of
the
dependencies.
So
then
I
posted
a
comment
that
maybe
our
import
boss
is
not
working
at
all.
I
need
to
check
this
actually
because
it
seems
to
me
like
that
this
particular
dependency
was
not
in
the
import
boss,
but
it
was
tolerated
still
being
present
in
the
source
code.
A
B
E
A
E
E
A
E
A
Because
it
feels
to
me
like
when
this
member
is
being
added
at
that
point,
it
doesn't
matter
if
we
pull
em
or
not.
We
wait
after
the
odd
member
command,
so
we're
basically
waiting
here
for
the
question
to
be
available.
Yeah,
okay,
I
would
also
do
this.
If
you
please,
please
take
a
look
if
you
came
all
right,
I
already.
A
A
What
was
wasn't
that
big
of
a
problem
in
just
the
cub
later
on
the
worker
reported
a
bunch
of
errors,
but
nowadays,
if
abilities,
the
worker
can
break
the
control
plane,
dogs
in
a
situated,
the
the
controller
components
start
restarting
constantly,
and
they
leak
memory
from
what
I'm
seeing
and
the
lake
memory.
At
some
point,
the
node
is
so
cloaked
that
its
payments,
the
company,
stopped
okay.
B
B
A
Sir,
basically
I
tried
to
explain
this.
To
that
it
may
depends
who
is
up
the
operator
of
the
cluster,
because
you
can
have
an
admin
that
has
the
ultimate
privilege,
but
other
users
can
be
given
tokens
like
in
the
case
of
Canadian.
They
can
have
both
served
audience
and
some
users
can
perform
as
many
joints
as
they
want
and
they
can
make
this
mistake
and
they're
like
small
implication
of
attack
vectors
here,
but
the
Jordan
is
basically
delegating
this
for
the
both
rappers.
A
A
So
my
PR
is
a
bit
problematic,
because
I
forgot
to
think
about
the
scenario
when
you
join
a
worker,
it
fails,
and
then
you
try
to
rejoin
the
worker
this.
My
code
is
currently
going
to
block
this,
which
is
not
desirable.
Also
the
node
restart.
The
question
is
going
to
keep
the
law
does
not
ready.
You
will
start
the
load.
You
try
to
join
it
again
and
again.
A
B
A
Yeah,
if
it's
ready,
especially
if
we
are
seeing
this
like
major
breakage
in
upstream
kubernetes,
this
definitely
started
happening
some
somewhere
between
maybe
somewhere
before
the
support
scheme.
It
is
something
like
110
111,
because
I
always
join
nodes
by
accident,
and
this
wasn't
the
case
before
nowadays.
I
just
see
the
control
plane,
pods
constantly
restart,
which
is
very.
E
D
D
E
A
A
So
this
an
actual
bug
so
during
up
greatly
perform
preflight
checks,
one
of
them
is
to
check
if
I
mean
it's
not
correct
to
do
this,
but
we
check
if
the
node
the
control
plane
node
is
tainted
and
in
single
control
based
scenarios,
the
user
does
not
taint
the
node,
so
you
know
with
the
master
taint.
So
this
basically
prefer
Jake
currently
fails
for
single
control.
Brain
scenarios,
which
is
some
people,
do
it-
is
like
a
mini
cube
but
using
key
medium.
So
I
think
this.
A
E
A
A
B
F
A
A
So
this
is
a
ticket
that
somebody
from
I
believe
from
the
testing
for
folks
created
that
I
need
to
update
the
documentation
for
this
grid
changes
so
super
important
for
now
those
export
we
are
working
on
that
this
were
passing
with
the
privileged
user,
so
yeah
I
was
able
to
first
I,
wasn't
able
to
reproduce
it,
but
then
I
realized
that
the
problem
is
actually
in
the
integration
tests.
It's
not
in
the
unit
test.
So
if
you
put
this
command
so
cube
ATM,
there
is
a
folder.
A
If
you
go
to
the
kkc
medium,
there
is
this
test.
Folder
and
the
dis
former
contains
the
integration
tests
for
Canadian
the.
We
call
them
integration
tests
because
they
require
you
to
have
the
binary
and
you
feed
the
binary
with
some
flags.
And
then
you
expect
results
like
is
this
going
to
succeed?
Is
this
going
to
fail?
And
basically,
if
you
look
at
the
regularly
on
tests
that
are
called
use,
you
know
with
go
go
test
and
the
qadi
namespace.
E
E
A
A
E
A
A
A
A
Let
me
see
if
I
apply
this
yes,
so
they
claim
that
this
support,
but
it's
not
really
a
body
just
something
on
their
system.
So
I
do
the
/,
sorry
/
3
hours
report
and
then
remove
the
kind
book.
If
something
is
actually
important
for
kube
a
game.
What
I
do
is
I
apply
the
/re,
a
cube,
a
DM
label
if
there
is
already
suspicion
on
the
ticket,
but
if
there
is
no
discussion,
I
ask
them.
A
Ok,
I,
usually
I,
do
a
lot
of
triage
in
this
repo.
Something
like
that,
I
don't
do
is
watch
the
qadian
channel
because
I'm
very
overbooked
already
I
think
for
breach
generosity
of
doing
that.
But
yeah
I'm
going
to
ask
you
if
something
is
somebody's,
not
moving
their
ticket
I'm
going
to
ask
you
for
that:
ok,
yeah!
So
if
they
might
has
time
you
can
occasionally
check,
you
know,
look
for
the
cube,
a
DM
label
and
try
to
help
me
with
that.
A
The
way
to
look
for
the
cube,
alien,
PRC's,
pretty
simple,
you
just
have
to
filter
a
filter
by
the
area,
cube
a
DM
level
that
is
added
automatically
if
you
touch
something
in
the
cube,
a
DM
folder.
So
this
is
from
Ed
bartosz.
We
spoke
Paris.
He
is
trying
to
remove
a
stray
dependency,
so
this
I
saw
it
today,
but
they
didn't
have
have
a
deeper
look.
A
A
A
B
A
B
There
was
a
problem
that
happened
with
the
the
proxy
and
that's
that
goes
against
policy
right.
You
can't
just
add
to
a
version,
and
we
got
ourselves
in
a
major
snafu
for
an
upgrade
scenario.
That
I
happened
around
that
the
110
timeframe,
because
there
was
no
ability
to
do
conversion
across
the
previous
renditions.
So
it's
totally
violates
policy.
B
A
B
So
there
is
a
weirdness
of
adding
things
to
be
in
one
fields,
for
this
considered.
Ok,
under
the
scenario
that
it's
omit
empty,
like
what's
really
weird,
is
like.
We've
documented
some
of
this,
but
I
have
not
seen
official
policy
and
field
editions
for
un
objects
right
like
every
time,
I
try
to
nail
this
stuff
down.
It's
like
we
start
talking
about
like
what
does
the
color
blue
taste
like,
and
we
just
run
in
circles
at
some
point,
we
really
need
to
I,
don't
know
what
it's
gonna
come
I
think
we
need
to
finish
off.
A
A
A
A
So
this
this
is
a
peer
I
reviewed.
It
touches
comedian
because
it
is
doing
a
factor
of
our
logic
that
we
have
about
splitting
Yama
documents
and
I,
don't
think
the
user.
Sorry,
the
contributor
is
working
based
on
some
discussion
or
a
ticket
which
is
I'm
not
sure
what
is
going
on
there,
so
I'm
going
to
assign
working
group
component
standard
folks,
please,
let
me
see,
frost
is
Orion.
This
I.
F
G
Is
just
to
break
out
the
functionality
into
runtime?
Essentially,
so
the
this
is
required
for
multi
doc,
parsing
and
component
standard
needs
to
be
able
to
build
additional
libraries
in
component
base
flex.
Lumiere
suggested
from
this
curve.
Idiom
isn't
the
only
tool
that
needs
to
parse
multi
doc
things
so
Lucas
originally
had
a
PR
to
rip
this
out
and
Alex
is
the
owner
of
this?
Carry.
A
G
A
This
was
the
metal
factory
PR.
This
is
subtle.
Ice
also
checks
a
long
time
ago,
locked
the
ticket
where
he
said
that
we
use
the
the
errors,
are
up
functions
in
a
lot
of
places,
but
we
never
print
a
stack
trace.
So
one
of
the
proposals
was
to
maybe
print
a
stark
trace
for
developers.
So
I
have
a
look
at
the
basically
that
document
that
kubernetes
has
in
the
community
repo,
and
this
document
is.
Let
me
show
you
quickly
it's
about
listing
the
different
like
eliminating.
What
are
the?
A
What
are
the
meanings
for
the
verbosity
levels
of
k
lock
before
that
walk,
so
I
looked
at
this
and
I
say:
I
saw
that
become
a
trace
level
verbosity.
So
a
stack
trace
seems
like
a
trace,
so
I
send
a
PR
basically
to
make
it
such
make
dinner
when
they
define
away
the
key
Bailey
makes
it
with
an
error
which
we
grabbed.
This
error
check.
If
the
guac
there
was
K
walk,
there
was
still
a
risk,
I
love
and
we
dumped
a
stack
trace
of
the
error.
A
This
means
that
if
somebody
is
working
on
Cuba
diem,
hey
I
want
to
see
the
stack
trace
of
the
error.
I
got
a
member
where
this
machine
is
coming
from.
I
can
do
this
and
it
is
going
to
show
me
the
list
of
functions
that
are
being
called
like
one
question
I
had
here.
Is
it
if
Yahoo
responded
with
+1?
If
do
we
want
to
always
show
The
Fault
message
like
hey
to
see
a
stack
trace
of
this
error,
execute
with
v5
I
want
to
get
feedback
on
this?
We
are
in
this
meeting
I
like.
B
That,
if
there
is
an
error,
yes,
if
there
is
an
error-
and
you
want
to
see
for
more
details,
run
run
this
command
with
-5
and
I
would
say
like
not
just
that,
but
you
should
rerun
the
exact
same
command
like
give
the
output
or,
as
the
output
of
this
basically
give
them
a
copy-paste,
full
entirety.
Statement.
A
B
D
A
So
this
is
going
to
catch.
Let
let
me
show
you
the
pier.
Basically,
for
instance,
we
had
K
walk
Sato.
Obviously
this
is
not
going
to
catch
it,
but
if
we
propagate
bubble
up,
this
error
is
going
to
eventually
go
into
something
like,
for
instance
like
this
one.
This
is
in
the
main,
cube
ADM
file,
and
we,
if
we
go
this
check
here
or
we're
going
to
catch
a
stack
trace
but
you're
talking
about
goroutines,
erase
conditions
and
things
like
that,
correct
we
there
is
no
way
to
because.
A
E
E
So
when
we
really
run
it
with
with
v5
we're
not
going
to
see
the
same
error,
maybe
we
could
create
a
temporary
file
when
we
have
back
trace
on
there
temp
and
we
say:
ok,
this
key
medium
crust
and
you
have
directories
here
on
TMP
without
the
need
of
providing
v5
for
more
than
that.
But
I
mean
this
is
fine
cities
for
me
as
well.
I.
B
A
E
A
Yes,
it
is
definitely
going
to
mix,
but
it's
already
mixing
and
there
has
to
be
a
way,
but
that's
one
of
the
reasons
I'm
not
in
it
only
are
for
a
very
high
velocity
level.
So
when
you
do
machine,
readable
output
reminder
setting
everything
over
a
v-0
is
going
to
be,
which
is
going
to
interrupt
the
machinery
about,
put
everything
that
is
more
than
v-0,
so
every
K
walk
call
is
going
to
be,
is
going
to
interrupt
the
output
of
the
in
the
terminal,
yep.
A
Think,
to
be
honest,
I
think
we
should.
We
can
add
this
and
I
would
like
to
see
the
machine
readable
to
solve
some
of
these
problems,
because
we
in
a
lot
of
places
week,
okay,
walk
it
entered
standard
output
with
an
error,
but
then,
if
the
the
user
has
typed
the
standard
output
to
a
file
to
you
know
accepting
some
solid
Jason
out
or
something
like
that.
It's
fine!
But
this.
A
If
we
have
other
messages
over
the
place,
that
output
is
something
that
is
not
part
of
the
start
of
the
Machine
revolt
put
in
a
separate
goroutine,
then
we
have
a
concurrency
problem
with
pollution
of
the
machine.
Readable
put
so
it's
it's
a
very
interesting
topic
that
I'm
hoping
that
mushy
read
more
effort
is
going
to
solve
this,
but
probably
not
for
the
cycle.
B
In
a
separate
location
seems
okay,
especially
if
it's
temp
the
I
don't
know
I
have
to
think
about
it.
I
think
we
should
do
it
here.
I
think
it
requires
a
little
more
thought
because
it
might
be
slightly
nuanced,
because
it's
entirely
possible
that
there
could
be.
You
can
just
put
it
in
temp,
because
if
you
know,
if
there's
potential
security
constraints
around
what
you
actually
put
command,
you
are
running
so
I
need
to
think
about
it.
Yeah.
A
A
This
is
a
chair
yes
discussed.
This
is
a.
This
is
a
technically
jariabek
where
we
forgot
to
update
the
canadian
constants
for
control,
laying
current
registration
minimum
complete
version,
and
we
have
to
create
a
ticket
to
automate
this
roster.
You
want
to
create
a
ticketing
communities,
palladium
phonus,
yeah,
sure,
okay,
thanks
because
I
want
to
forget
again,
at
least
for
116,
we
already
updated
them,
which
is
good.