►
From YouTube: 2019 01 23 19 07 28 kubeadm office hours 179916854
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
Okay,
hello,
everybody:
this
is
the
sequester
lifecycle,
cue
medium
office
hours
of
23rd,
of
January
2019.
We
don't
have
any
sick
chairs,
but
we
have
some
agenda
items
to
discuss
today.
Please
all
your
names,
the
list
of
attendees
and
add
your
agenda
items.
If
you
have
any
for
Brito.
You
have
the
first
item.
B
Yes,
go
everybody,
so
the
first
ballot
item
regards
some
discussion
and
some
issue.
Some
discussions,
like
a
semi
issue
on
the
covalent,
mean
repository
about
what
what
is
supported,
what
what
is
officially
supported
by
cubed
mean
and
what
is
not,
and
we
Furr
guards
to
the
cluster
variant
that
you
can
achieve
using
phases
and,
in
my
opinion,
now
that
the
the
results
to
the
skip
feces
flag,
the
users
starting
use
it
consuming
using
it.
And
but,
in
my
opinion,
when
you
start
using
physics,
you
are
not.
B
B
B
Which
is
the
the
first
point,
and
second
point
is:
if
do
we
want
to
increase
the
number
of
cluster
variants
that
are
supported,
especially
with
regard
to
the
keys
of
standing
a
cluster
without
proxy,
because
there
is
out
there
are
other
tool
like
Google
Earth
that
can
replace
it.
I
don't
know
if
there
are
opinion,
we
forgot
to
dis.
C
B
Have
for
me,
a
clusters.
Variant
is
a
cluster
that
use
an
externality
CD
or
a
cluster
that
use
cubed
ENS
instead
of
core
B&S.
Those
are
two
clusters
variants
which
are
fully
supported,
but
coogan
mean
because
we
check
now
manually
but
in
futuro,
I
hope
with
CI
CD.
We
check
out
every
release
that
all
the
work
or
the
kubja
meaning
work
for
works
for
this
variant,
but
it
is
not
feasible
to
do
these
for
all
the
possible
combination
of
phases
or
whatever.
C
Yeah
I
see
what
you're
saying
so
I
would
say
that
the
it
sounds
like
we
do
have
variants
of
clusters
that
we
do
officially
support.
It
would
always
be
good
to
document
that
and
make
that
explicit
and
then
adding
adding
some
kind
of
some
kind
of
warning,
like
two
phases
at
least
letting
people
know
that
they
may
create
a
cluster.
Some
sort
of
Franken
cluster
that
we
don't
support,
would
definitely
be
a
good
thing
to
do.
B
It
is
taught
in
a
gene
than
now,
but
the
the
question
is
that,
okay,
we
know
more
or
less.
We
discussed
now
that,
if
you
use
faces
you
are
know,
you
are
going
out
from
the
official
support
or
from
the
safe
path.
And
the
next
question
is
that
okay,
we
want
do
we
want
to
even
a
new
clusters
variant
that
is
a
cluster
or
without
who
the
proxy
that
that's
a
question.
If
there
is
consensus,
it's
something
that
we
have
to
look
for
volunteers
and
to
plan
for
it.
Yeah.
D
It's
something
it's
something
very
wide
I
mean,
but
I
think
that
integration
part
is
a
good
thing,
because
why
maintain
many
things
you
can
just
integrate
of
something
that
works.
Let's
take
as
may
be.
Cows
is
an
example,
but
this
is
just
one
of
many
examples.
I
think
out
there
in
the
list,
so
but
I
think
it's
good
idea.
A
A
B
Don't
see
a
problem
in
that
because
if
the
user,
so
there
are
things
that
are
in
charge
of
kuben,
mean
and
things
that
are
managed
by
the
user.
So
if
the
user
decided
to
do
the
cluster
without
the
proxy,
then
the
proxy
solution,
the
user
we
distal
is
up
to
him.
So
it
also
it
goes
out
from
the
scope
of
cookie
the
mean
is
up
to
the
user,
a
starter
and
up
greater,
so
I'm,
not
I'm,
not
worried
about
what
goes
out
from
the
scope.
B
E
E
B
D
D
B
Yes,
but
ok,
we
are
I,
think
that
what
we
are
saying
is
that
it
makes
sense
to
support
this
scenario
broth,
but
from
saying
that,
and
from
finding
someone
that
does
these
implemented
these
and
figure
out
all
the
detail.
We
need
some
volunteers
I
think
that
for
this
for
this
cycle
there
is
no
room
unless
someone
wants
to
take
they
take.
This
is
probably
the
different
item
to
solve.
I
would
open
an
issue.
I
will
simply
simply
open
an
issue
and
and
say:
ok,
guys
that
are
there
are
requests
about
these.
A
Maybe
we
could
reuse
the
old
ones,
so,
let's
just
rename
the
title.
Yes,
it's
your
choice.
My
question
is
so:
where
should
we
document
it
because
I'm
kind
of
confused,
which
page
in
the
docs
we
should
use
for
stuff
like
that,
like
I
I,
have
to
check
yeah
I
have
no
idea,
probably
somewhere
in
queue
baby
a
minute.
There
is
a
paragraph
about
phases.
We
should
probably
mention
there
like
a
caution,
paragraph.
Okay,
if
you
use
custom
phases,
you
are
on
your
own
in
terms
of
this
coastal
cluster
and
abrasive
work
for
now,.
B
A
B
A
B
Yes,
but
but
I
I'm,
quite
convinced
that
now
we
will
fix
the
joint
part
during
the
real
work
of
phases.
The
problem
there
is
that
we
are
reusing
the
Dame
we
are
reading,
the
fully
need
configuration-
and
this
implies
also
read
all
the
component
configure,
but
this
is
not
necessary
in
when
doing
join.
A
B
B
There
is
us
a
set
of
issue
that
are
popping
out
about
how
to
do
certain.
You
are
saturation
a
certain
link
at
and
so
on,
says
a
certain
is
a
really
a
complex
topic
and
we
already
did
some
some
work
in
trying
to
make
it
the
documentation
better,
but
it
seemed
that
this
is
not
not
enough,
so
I
want
only
to
add
a
PSA.
If
someone
has
no
knows
the
topic
whale
or
want
to
deep
dive
this
topic,
it
will
be
welcome.
A
So
this
is
the
next
person
this
I
also
saw
was
it
Raphael
from
sushi
also
spoke
in
the
search
I
personally
do
not
consider
myself
any
type
of
expert
in
search.
My
questionnaire
was
like.
We
already
have
a
document
that
I
mean
it's
not
super
detailed
way,
but
it
already
contains
some
guidelines
how
to
renew
certificates
and
stuff
like
that
with
qadian.
Do
you
see
users
complaining
about
this
document.
B
A
Yes
again,
this
is
one
of
those
things
that
it's
they
are
very
important,
but
we
like
you
and
me,
and
the
others
we
have
like
different
priorities,
the
cycle,
and
unless
someone
decides
to
help
us
I,
don't
see
how
we
can
fix
them,
because
the
release
is
going
to
be
late.
March
I
think
like
hold
freeze
is.
A
F
E
B
C
E
C
A
So
the
particular
problem
from
Angus
Angus
Lee,
the
first
issue,
I,
think
that's
one
of
why
I
haven't
seen
this
before
so
I'm,
going
to
consider
it
as
being
a
special
case.
I
mean
that's,
how
I
see
it,
but
we
probably
shouldn't
spend
time
now
and
just
at
the
help-wanted
labels
I
mean
that's.
That's.
A
The
help-wanted
never
without
it
there
so
do
you
know
if
this
is
going
to
be
back
to
cube
ADM
like
this
cycle
or
something
that
probably.
A
A
A
B
F
G
B
B
E
A
G
A
Understood
yeah,
or
is
it
possible
that
after
using
container
D
and
then
uninstall
Enoch,
the
socket
could
remain
that
like
present
on
the
distro,
and
could
that
create
a
problem
like
you
might
end
up
with
multiple
sockets
and,
like
you
say,
container,
they
might
not
be
old,
it
might
not
be
present,
but
the
socket
is
there.
I
can.
A
B
Okay,
then,
then-
and
there
are
some
old
PR
which
are
out
there
since
time-
that
it
will
be
great
to
discuss
together
and
see
if
we
can
reach
an
agreement
and
close
them.
And
maybe
we
I
don't
know
if
there
are
more
important
item
in
agenda
and
if
the
more
important
items
in
agenda
we
do
these
at
the
end
of
the
meeting.
So.
A
B
Okay,
so
I
gave
a
brief
introduction,
so
two
one
or
two
cycle
ago,
I
guess
I
forgot,
said
this
loss
it
that
makes
could
mean
search
for
the
Cooper
configure
in
a
similar
way
that
more
similar
way
then
kuba,
scuttle.
So
first
look
for
environment
are
bad
for
environment
variables,
then
from
the
India
Alma
and
and
finally,
it
looks
at
the
where
they
could
mean
expect
the
admission
be
the
problem.
The
problem
that
I
see
in
this.
B
So
I
see,
in
my
opinion,
all
th
and
that
manager
both
type
of
configuration,
so
local
and
cluster
should
not
rely
on
on
this
Cooper
config
search
logic,
but
should
rely
only
on
the
admin
duct
conf.
That
is,
you
know,
in
a
well-known
place.
Otherwise,
for
instance,
I
have
a
machine.
I
restarted,
Mini,
Cooper
I
ran
away.
Is
it
what
happened.
B
Or
I
ran
a
phase.
What
happened
that
the
the
phase
right
the
manifest
locally
and
then
created
a
backer
rules
or
the
config
map
in
some
other
cluster,
so
I
see
it
the
very
risky
and
there
was
and
I
remember
also,
you
should
talk
about
this
if
we
make
a
cubed
mean
searching
for
Cooper
config,
but
it
that
is
the
risk
that
could
mean
make
such
action
on
home
place.
Where
are
not
on
the
route
contra.
A
A
B
A
B
A
It's
a
zero
point
here.
We
should
definitely
present
this
in
front
of
team
as
well.
I
think
the
I
agree
with
you,
because
these
are
not
local
commands.
They
are
not
about
like
control
in
an
external
cost
somewhere
in
yes,
I
think
that
any
join
every
said
should
be
without
these
procedures
of
the
environment.
Very
great
yeah.
A
See
what
you're
saying
yes
yeah?
Unfortunately,
we
have
to
block
on
the
team
and
look
us
probably
on
this
because,
like
they
should
give
the
final
call,
but
yeah
definitely
for
the
rest
of
the
commands.
However,
like
this
PR
is
touching,
the
rest
of
the
command
post
will
be
possibly
reset
as
well.
Is
that
correct
I
have.
A
A
A
But
this
again
now
this
is
this
also
applies
to
keep
kind
of
in
general,
like
ICP
I've
seen
reports
from
people
like
messing
up
with
cupca,
no
because
keep
Carol
by
default
searches,
environment,
variables
and
they
forgot
to
set
it
to
the
the
new
cluster
they
created
and
it
basically
apply
stuff
to
the
odd
coaster.
So
that's
like
it
may
see
it's
messy.
You
cannot
be
like
super
dangerous,
but
it
can
be
very
messy.
C
C
A
B
This
is
a
part
of
init.
If
I
may
respond
to
Chuck
I
agree
it
Amit,
it
could
be
confusing
having
two
different
approach,
and
so
it's
something
that
we
should
consider,
but
I,
don't
think
that
it
is
a
problem
at
the
end
of
combat
meaning
it.
We
tell
a
user
if
you
want
to
use
Cooper
cut
or
copy
the
compa
file,
but
it
is
for
using
Cooper
cut
or
not
for
music
Cooper
Mina
again
I
see.
C
D
A
D
A
H
B
E
B
B
For
other
common
light
like
abroad,
config
like
token
those
comment
are,
are
utility
and
it
could
make
sense
to
eff
them
act
in
the
same
way
of
kata,
because
you
can
use
also
on
on
different
machine
that
in
the
node
are
part
of
your
cluster.
So
it
is
here,
it
makes
sense,
but
we
were
discussing
this
now
with
it
is
kind
of
confusing
I,
think
open,
meaning
using
the
two
logics
I
think.
B
A
Bridge
we're
saying
that
we
should
let
have
this
president's
order
for
any
joiner.
That
I
said
the
notes
in
advance,
basically
and
completely
saying
that
we
should
search
in
very
food,
qadian
path
first
and
then
like
search
environment
variables
and
for
the
rest
of
the
commands.
We
already
have
this
presence.
A
A
H
I
think
it
depends
upon
like
what
for
rates.
You
were
saying,
the
user
experience
for
different
operations
like
if
a
person
using
the
token
subcommands
independently,
like
using
standard
precedence,
order
of
coop
control,
makes
perfect
sense,
because
you'd
expect
that
behavior,
because
you're
acting
on
any
existing
cluster
I
think
it's
it's
totally
dependent
upon
whether
you're
acting
on
a
cluster
or
whether
or
not
you're
creating
a
new
cluster.
H
H
So
I
think
any
I
think
that
there's
those
that
the
rule
that
I
just
set
up
arbitrarily
in
my
head
actually
makes
sense
is
that
if
you're
acting
on
an
existing
cluster,
then
you
should
follow
a
default
precedence
order
if
you're
creating
a
cluster
in
any
way,
shape
or
form.
You
should
follow
the
order
that
Fabricio
had
mentioned
earlier,
but.
H
B
H
H
When
a
person
sets
down
to
do
a
reset
or
an
upgrade,
they
should
know
the
environment
that
they're
setting
up
in
and,
if
they're
trying
to
do
it
remotely.
You
know
they
should
specify
the
command
line
overrides
for
that
specific
environment
if
they
need
to
right.
Otherwise,
if
it's
local
environment,
they
should
expect
the
local
settings
right.
A
A
A
H
A
A
A
A
You
know
the
links
are
always
stable
and
at
some
point
they
stop
using
the
link
checker
and
now
it's
like
our
out
of
the
loop
completely,
but
still
we
generate
these
placeholder
documents
that
are
part
of
the
link
checker
basically,
and
the
placeholder
documents
like
according
to
Aaron
and
Fabrizio,
and
even
the
sick
dogs
people.
They
are
completely
useless
right
now.
We
just
have
to
convince
lava-lamp
and
I
I.
A
H
Just
seems
like
there's,
it
seems
like
one
of
those
ownerless
efforts.
It
doesn't
seem
like
there's
conflict
on
what
needs
to
be
done.
It's
just
the
question
of
like
is
someone
gonna
see
it
all
the
way
through
and
make
sure
that
on
the
change
it
doesn't
break
something
else
weird
I
can
I
could
poke
around
that
issues
been
open
forever
like
since
October
right
or
some
like
that,
yeah
and
the
reason
we
started
doing.
H
That
was
because
every
time
we
did
a
PR
to
change
something
we
were
generating
these
docs
when
we
did
a
command-line
change
and
it
was
just
it
required-
need
someone
else
who
had
administrative
Ackles
to
override
the
approval.
So
we
did,
you
know
we
didn't
sort
of
fall
into
the
trap.
There's
another
way
around
this
too,
which
is
just
to
give
us
approval
rights
on
the
docks.
It's
a
question
of
like
what's
faster
yeah
I
can
I
can
streamline
that,
and
we
just
get
through
that,
and
we
don't
have
to
team
at
that.
A
So
everyone
basically
said
that
he
doesn't
want
to
give
us
access
into
the
docs
folder,
because
the
dog
folder
also
contains
the
API
reference
documentation
and
we
do
not
own
that
yeah.
That's
what
that
was
the
argument
for
Marren
and
he
decided
that
we
should
go
to
the
next
route
and
figure
out
like.
Why
is
this
link?
Checker
placeholder
dog
still
present
there
and
I?
Guess
it's
only
a
matter
of
convincing
Daniel
Daniel
to
stamp
this
okay
yeah!
That's
it
all!.
A
Doesn't
sound
as
good,
alright,
alright
quickly,
we
have
another
another
last
one
and
it's
I
don't
think
we
can
fit
this
in
ten
minutes,
but
basically
it's
about
the
local
DNS,
cache
and
I
I
think
we
should
focus
like
everybody
can
drop
a
sentence
here.
Mike.
Do
we
want
to
add
this
into
the
next
cycle
this
cycle
and
what?
How
should
we
do
it?
Should
it
be
part
of
the
config?
Should
it
be
like
a
pivoting
function
in
now
for
somewhere?
Should
it
be
a
feature,
gate
and
yeah?
A
H
A
A
A
Something
beyond
Lucas
was
were
confused
about
the
we
have
to
use
this
new
container,
which
is
called
Akkad's
dns,
not
cache,
and
it's
maintained
by
the
same
people
that
maintain
to
be
honest
and
I
started
questioning
the
foxes.
If
dns
goes
away,
are
they
going
to
continue
maintaining
this
container
that
we
now
have
to
use
in
parallel
with
core
DNS,
so
Tim
Hawking
came
by
said:
hey,
we
are
going
to
continue.
F
H
Need
to
root
out
the
core.
This
issue
I
think
there.
There
are
different
DNS
deployments,
because
the
problem
here
is
like
how
we
used
to
do
things
in
the
past
and
OpenShift
versus
how
they're
currently
done
in
the
upstream
Kubb
up
implementation
versus
how
they're
currently
done
inside
of
the
coop
ATM
deployment.
H
There
was
where
the
local
DNS,
caching
and
the
local
resolution
of
the
resolver
were
first
set
as
daemon
sets
in
the
past
and
I.
Don't
know
exactly
how
folks
are
doing
this
currently
with
the
coop
deployment,
so
I'd
have
to
take
a
look
there,
but
it
seems
weird
that
this
would
be
part
of
a
deployment
of
core
DNS
right
like
or
you
can't
do
it
as
what's
inside.
This
was
different
inside
of
open
shippers.
We
had
something
called
a
deployment
configure.
H
E
A
H
Possibly
I
wanted
to
do
a
quick
drop
of
the
status
of
a
packaging
conversation.
Yes
please!
So
it's
it's
just
confusion.
Currently
there
was
Hennis
is
working
from
the
release.
Perspective.
I
didn't
know
that
we
had
a
Capezio
item
to
fix
packaging
in
KK
and
that's
what
that
main
issue
is
meant
to
solve
and
I
know.
I,
don't
know
if
Marik's
on
I
know
folks
from
sue
say
had
decided
that
they
were
going
to
help
on
that
effort,
and
there
were
other
folks
who
are
also
going
to
help
on
that.
H
I
know
that
even
some
except
do
folks
were
also
helping
drive
some
of
that
conversation,
but
that
is
still
a
p0
and
you
still
should
work
on
and
deliver
that
issue,
which
is
basically
building
from
basil,
builds
in
kk,
basil
or
otherwise.
The
ability
to
Jennifer
ate
the
Deb's
and
rpms
the
canonical.
What
we
consider
to
be
canonical
versions
of
those
artifacts,
so
that
should
still
be
p0
folks,
should
still
execute
as
planned.
H
Some
of
tests,
infra
depending
upon
where
you
are
what
jobs
are
running,
will
consume
the
nightly
artifacts
and
some
of
the
some
of
them
will
actually
consume
the
stable
artifacts
right,
so
you'd
have
to
you'd
have
to
fix
all
three
problems
and
they're
all
separate.
Well,
the
publishing
and
some
of
the
testing
are
kind
of
linked
together,
the
the
canonical
artifact
generation
we
can
fix
in
KK
like
today.
There's
nothing
blocking
us
from
doing
that,
and
we
should
still
continue
doing
that.