►
From YouTube: Kubernetes Federation WG sync 20180702
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).
B
Hello,
you,
oh
so
Paul
I,
don't
know
if
you
notice
I,
ask
a
question
on
an
issue,
but
probably
should
just
ask
you
directly
when
you're
implementing
support
for
status,
sub
resource
did
I'm,
pretty
sure
you
didn't
just
double-check.
Did
you
add
support
for
installer
generation
for
that
status
of
resource
as
well.
C
B
It's
a
bug,
there's
kind
of
two
paths
and
I
would
suggest
that
the
install
C
or
these
option
or
the
controller
is
really
a
test.
Only
thing
the
hack
install
but
yeah
mole
that
Q
builder
generates
allows
you
to
like
round-trip,
so
you
can
be
like
cube
cattle
apply
installed
on
ammo
and
can
also
do
Q
kettle
delete,
install,
buy
em.
All
you
can't
do
with
the
controller
variant.
It
only
will
allow
you
to
install
things
now.
C
I
tested
it
with
sterilizing,
the
CR
di
mo
and
I
know
that
it
works
they're,
like
the
tests
established
that
if
you
have
the
status
of
resource
annotation
on
a
on
on
like
in
the
code
in
the
right
place
that
the
generated
CRD
a
mole
has
the
the
status
sub
resource
in
it.
But
it
might
be
that
whatever
serialize
is
install
that
llamo
might
use
like
a
slightly
different
pass.
So
we
can,
we
can
look
into
it.
It
should
be
present,
though.
So
if
it's
not
at
the
bug,
yeah.
B
So
that's
kind
of
where
I'm
at
I
did
manage
to
get
local
a
cluster
working
to
some
degree,
but
then
I
ran
into
the
fact
that
local
cluster
uses
local
hosts
and
when
you
try
to
join
a
cluster,
that
is
local
host.
That
predictably
doesn't
work
so
I
guess
I,
probably
I'm
sure
local
cluster
as
a
way
of
specifying
an
actual
IP
address,
or
name
or
something,
so
it
would
be
like
accessible
outside
of
the
container.
So
you
try
to
like
hey.
B
A
B
B
D
B
A
That
we
also
ran
into
that
status
related
problem
in
the
room.
Then
we
were
trying
to
figure
out
so
steel
wrapping
our
heads
around
that
so
are
the
management
which
is
like
integration,
fixture
prata
and
the
atrium
brought
up
by
the
tests
itself.
That
works
fine,
the
other
one
that's
a
task
for
us
is,
we
are
still
going,
I
mean
no
but
not
completed
yet.
B
Posted
kind
of
rudimentary
updates
to
the
deployments
it
is
a
little
bit
broken
like
the
deployment
scripts,
alright
and
it
it's
kind
of
you
can
manually,
get
it
working
if
you
edit,
the
install
IMO
but
I'm
so
I'm
working
as
you
probably
guessed
on
trying
to
fix
the
install
ammo
generation.
The
trick
is
just
that,
if,
like
the
way
I'm
doing
it
is
kind
of
like
I
generate
it,
then
I
go
edit,
yes,
but
that
it
just
isn't
a
very
seamless
experience.
B
If
dr.
dark
requester
is
working
for
you,
maybe
I
can
work
with
you
directly
to
just
verify
like
you
might
have,
because
you're
not
gonna.
Have
this
localhost
issue
that
I'm
having
we
might
actually
be
able
to
make
sure
that
the
end-to-end
is
running
a
darker
and
darker
for
you
I'm,
happy
to
coordinate
with
you
after
this
column,
not
sure.
A
Sure
and
I'm
not
sure
I
did
understand
it.
Clearly
the
issue-
and
you
didn't
explain
it
last
time-
the
issue
with
the
docker
version
version
you
were
talking
about
and
I
and
Shashi.
Both
of
us
actually
haven't
synced
on
what
the
docker
versions
we
are
using
or
not
I,
remember
last
I
updated.
It
was
many
months
ago,
maybe
six
months
ago
or
so
and
I
have
bid
update
the
doc
aversion
to
some
of
the
community
versions.
I
mean
they
have
now.
Enterprise,
darshan
and
committee
version
right.
So.
A
B
As
far
as
I
can
tell
that's,
kind
of
the
stumbling
block
like
I
did
have
one
probably
more
fixed
to
post,
like
I
haven't
reached,
but
related
to
defaulting
the
federated
type
configs
that
are
defined,
don't
actually
they're,
not
complete,
because
I
was
assuming
default
and
I'm
actually
setting
the
default
and
the
federated
type
config
controller
before
it
launches
st.
controllers.
It'll
just
it'll
call
the
little
set
defaults
on
on
the
configs,
and
that
seems
to
work.
I
can
do
in
memory.
B
Controllers
against,
like
a
cluster,
would
suggest
that
we're
not
seeing
control,
plane,
conflict
or
interaction.
So
the
only
real
thing
is
just
like
doing
an
end
and
making
sure
it
works,
but
I'm
pretty
confident
that,
like
if
there's
any
issues,
it's
not
any
conflict
with
cube.
It's
just
us
getting
things
installed
properly,
which
is
to
me,
that's
that's
low
risk.
We
just
have
to
go
through
the
motions.
A
C
In
terms
of
like
practical,
practical
things
that
we
need
to
work
through
before
we
can
release
an
alpha,
I
think
we
still
have
the
same
concerns
that
we
identified
last
week.
Right
like
we
want
to.
We
want
folks
to
test
Marissa
poll
and
make
sure
that
it
can
that
the
Q
Builder
refactor
works
correctly,
with
a
real
with
real
kubernetes
clusters
right
and
then
there's
also
like
Maru.
C
You
wanted
to
make
sure
that
Shashi
and
our
farm
were
ok
with
changes
that
you
made
to
the
features
that
they
worked
on
and
I
think
also
what
I'm,
what
what
we're
kind
of
circling
right
now
to
like
put
a
really
fine
point
on
it
is
that
we
need
to
understand
which,
which
docker
versions
and
like
which
versions
of
dependencies.
You
need
right,
I,
think.
B
A
And
I
don't
think
this
necessarily
is
a
blocker
for
us,
so
we
can
always
have
this
comment
and
think
whatever
they
treat
me,
we
are
going
to
update
that
this
is
sort
of
a
prerequisite
and
I,
don't
think
is
too
much
of
an
ask.
Okay,.
C
B
So
speaking
of
that
I'm
wondering
wondering
if
you
can
look
at
what,
like
the
way
that
you
builder
builds
images
and
retarget,
so
stable
image
build
to
that
even
like
using
the
pr
pre-merger
just
because
I
mean.
Ideally,
we
have
a
sense
of
like
what
the
end
user
experience
would
be,
and
my
hope
would
be.
It
would
be.
You
know,
here's
a
URL
for
the
install
the
mo
because
we
would
commit
it.
B
Maybe
we
would
we'd
call
it
hack,
install
stable
so
that
you
can
still
generate
like
one
from
the
source,
but
then
using
that
would
refer
to
the
stable
image.
So
we
would,
you
know,
build,
however
often
and
so
you'd
be
able
to
just
like
keep
apply.
Url
I,
don't
know
if
you
can
even
do
that,
maybe
have
to
download
first,
but.
C
B
Well,
as
a
developer,
I
mean
you
do
that
pretty
pervasively
nobodies
I'm
not
suggesting
anyone
do
that
on
a
production
cluster
but
yeah
the
tires
do
this
in
a
VM
and
you're,
pretty
safe,
yeah
yeah
and
like
on
example
like
if
you
do
this
on
mini
cube,
you're
gonna
get
something
that's
already
contained.
Yeah.
C
B
C
C
B
C
B
B
C
B
A
A
B
Yeah,
well,
why
don't
you?
Is
it
like?
How
early
do
you
want
to
me?
I
mean
well,
I,
guess
I
could
do
legwork
beforehand
and
make
sure
we
have
everything
ready
and
then,
if
I
mean,
if
you
could
just
get
it
working
great
and
if
you
can't
but
I'd
like
to
work
with
you,
so
we
can
just
speed
it
along
well.
Tomorrow
is.
A
I
also
wanted
that
they
like,
let
me
try
it
for
a
day.
Maybe
I
could
just
get
it
working
on
my
own
in
case,
there's
a
well,
then
we
can
sync
up
tomorrow.
So
as
early
as
like
I
think
you
you
generally
do
early
mornings,
so
8
or
9
morning
is
perfectly
fine
for
me
that
you
give
us
a
couple
of
before
the
next
tomorrow's
meeting
awesome.
So.
B
Also
just
that,
if
you
get
it
working
like
if
I
manage
to
get
things
and
it's
gonna
stay
where
you
can
just
reproduce
it
great,
if
not
send
me
a
meeting,
invite
for
8:00
a.m.
and
otherwise
I'll
see
you
at
10:30
meeting.
Sure
ok
sounds
good
I'm,
just
trying
to
get
like
things
in
a
shape
where
I'm
not
walking,
because
I'm
gonna
be
off
Wednesday,
Thursday
Friday
so
today
and
tomorrow
is.
Is
it
for
me
this
week
and
I
want
to
make
sure
that
you
guys
can
can
move
forward
without
me.
B
B
One
quick
thing
to
mention-
and
this
is
for
you
Shashi
when
I
was
initially
I-
was
trying
to
make
sure
that
the
group
for
the
multi
cluster,
Resource,
remotes,
multiplicity
and
s
resource
was
multi.
Cluster
DNS
Kate's
thought
I
owe
you
builder
was
giving
me
a
lot
of
trouble
like
I
tried
like
manually,
fixing
it
and
it
just
there's
enough
disjoint.
It
really
expects
that
a
given
project
as
a
single
domain.
E
Yeah
I
just
solve
those
changes.
You
did
today
yeah
those
few
issues,
I
also
faced
I,
think
most
of
you.
If
you
already
solved
it
right
and
I,
think
the
path
is
mainly
the
control
manager
running
standalone
like
in
non-
cluster
right,
so
I
think
that
I
think
it
is
still
to
be
verified
that
the
talk--I
file
and
the
how
do
they
start
that
AK
install
dot,
yeah,
Mel
right
so
I'm
actually
facing
few
issues
with
so
once
I
install
Federation
I
joined
the
cluster,
but
the
cluster
is
never
going
to
a
ready
state.
So.
E
B
B
B
E
B
B
It's
not
just
like
validation
from
queue
builder
is
just
failing,
cube
111
like
it
was
fine
in
110
and
so
for
now
I'd
say
just
skip
it.
It's
basically
harmless
for
our
purposes.
I
mean
there's
consequences
to
including
filled
in
in
a
CR
D
which
are
compatible
with
sub
resources,
but
we're
not
doing
that.
B
E
A
And
similar,
so
there
is
this:
the
CR
that
CRE
definitions
are
in
API
extensions,
API
server,
and
that
version
also
is
pretty
old,
like
it's
1.1
0,
whereas
it
should
be
coherent
with
with
the
API
server
release
or
that
create
STDs,
which
is
a
newer
version.
That
is
the
reason
we
are
seeing
this
you.
B
Right
yeah
I
mean
I'm,
to
be
honest,
I'm
not
too
worried
about
that
at
the
moment,
because
we
can
work
around
it,
but
I
do
think.
The
four
alpha
will
want
to
have
the
newest
version.
I'm,
just
kind
of
like
I
mean
moving
to
a
different
version
and
worrying
about
the
consequences
or
something
I'd.
Rather
do
once
this
PR
emerges
and
yeah
or
media
less.
A
B
B
A
B
It's
kind
of
a
separate
path,
it's
looking
at
like
it,
it
doesn't
actually
read
the
type
so
like
everything.
You've
done
is
completely
fine.
It's
just
like
there's
another
path
path.
It
just
doesn't
implement
it,
yeah
so
I
gotcha
and
if
you
have
time
to
prioritize
that-
and
maybe
I
shouldn't
work
on
it.
But
to
me
this
is
like
a
blocker
that
needs
to
be
done
before
we
can
have
an
install
working
I'm
happy
to
continue
working
on
it.
What
would
you
prefer.
C
B
C
B
D
A
A
D
B
E
A
F
It
probably
is,
and
probably
has
all
the
versions
across
your
registry
that
it's
potentially
using
also.
This
is
my
first
stay
back
after
being
out
on
PTO
for
a
couple
weeks,
so
I'm
gonna
be
kind
of
looking
I'll
probably
spend
time
looking
at
that
eventually,
but
prioritizing
the
PR
that
Moreau
has
up
right
now.