►
From YouTube: Kubernetes Federation WG sync 20180703
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
Api
server
and
1.9
mini
Q
clusters
and
I
added
face
time
outs
couple
of
times,
but
that
was
I
guess
mainly
because
my
local
clusters
were
probably
would
not
get
the
pots
ready
in
time.
So
they
basically
first
time
if
it
is
run
than
it
might
download
the
images
for
the
deployment
pods,
so
otherwise
all
the
atria
tests
would
pass,
and
because
of
our
two
clusters
and
in
the
Federation
the
CRD
test,
there
is
a
test
for
CR.
B
C
B
Was
just
when
I
was
working
with
the
dynamic
stuff,
I
wanted
to
make
sure
that
we'd
be
able
to
federates
here
at
ease
and
we
can,
but
it
doesn't
really
require
like
multiple
clusters,
because
there's
not
really
any
behavior,
you
would
expect
from
a
C
or
D.
That
would
demand
that
it's
just
like
can
I
think
at
all.
That
was
it.
A
Okay,
yeah
and
I
spent
some
time
going
through
the
commits,
like
this
was
in
the
afternoon
when
I
did
when
they
eaten
after
that,
I
was
going
to
the
comics.
I
have
not
been
able
to
go
through
all
of
them,
but
I
guess
it
should
be
fine
to
merge
it
as
of
now,
because
there
are
many
places
which
are
to
do's
and
all
that
stuff
which
basically
either
you
or
one
of
us,
can
take
up
after
post
appear
and
post
this.
A
A
A
B
It
depends
on
whether
the
behavior
that
you're
testing
requires
me
to
eat
or
not
if
it's
something
that
reality
is
in
integration,
I
mean
at
the
very
least,
things
are
compiling,
so,
like
all
of
the
controllers,
have
been
updated
to
use
the
new
clients
a
new
and
the
updated
types.
So
unless
there's
specific
behavior
that
you
think
would
be
affected,
I
think
we
should
be
good
to
go,
but
I
mean
I,
encourage
you
to
double-check
my
assumption.
I.
A
B
D
B
C
B
C
B
A
Yeah,
that's
what
I
mentioned
so
I
actually
had
three
clusters
and
I
did
not
join
the
dinner
itself
because
it
had
some
like
it.
The
API
server
is
exposed
on
a
local
host
port.
So
if
I
join
it,
then
the
controller
is
not
able
to
reach
it.
I
did
not
bother
to
update
it
and
figure
out
how
to
get
a
public
IP.
So
what
I
did
was
I
had
a
dental
Esther
where
I
would
install
the
CR
DS,
and
that
would
be
the
Perdition
control
claim
and
two
different
mini
cube
clusters.
A
A
B
B
Cool
yeah
I'm,
so
I
have
created
like
a
script,
that
kind
of
works
around
the
issue
with
llamo
generation,
and
it
points
to
a
an
image
that
I've
pushed
to
darker
hub,
and
so
once
you
get
that
what
did
he
call
it
canary
built
once
you
get
a
canary
build
up
and
pushing
the
clay
there
key
that
we
can
update
that
the
scripts
are
the
points
to
something
other
than
an
image
and
I'm
just
randomly
pushed
yeah.
So
once.
A
D
B
A
D
Yeah
it
did,
it
did
an
image
that
was
a
really
simple
bill.
That
just
adds
the
API
server
and
controller
manager
binaries
into
like
Ubuntu
is
the
base
image
and
I
am
saying
that
we
have
so.
We
have
to
make
sure
that
that
build
works
correctly.
Now
that
we
don't
have
the
API
server
binary,
we
can
do
this
as
a
follow
up
like
immediately
after
Murray's
whole
merges,
but
good,
adapt
that
image
build
to
the
you
build
Erskine,
you
that's
all
yeah.
A
D
A
B
D
A
B
A
B
D
D
B
D
B
A
B
D
B
A
Yeah
that
should
be
fine,
I
I
noticed
one
more
thing
like
I
did
not
understand
the
reason
behind
that
it
might
be
some
problem
with
the
cluster
industry
and
it
might
not
be
blocker
for
our
alpha,
like
I.
Remember
that
earlier,
when
we
would
join
thrusters
and
add
the
twisters
to
crush
the
registry,
I
can
basically
query
the
cluster
resource
from
facilities
gain
I
would
get
that
these
are
the
drying
crushers.
A
A
B
A
B
B
B
A
B
D
B
D
B
B
A
B
B
D
B
C
D
B
Like
it's
something
like
the
issue
that
I
like
to
various
places
where
there
they're
fixing
the
validation,
it's
it's
gonna
land
in
112
and
hopefully
the
back
port
111
I,
don't
know
how
successful
will
be
getting
the
backboards
110
but
I'm
sure
we
can
ask.
But
at
this
point
the
manual
fixes
make
it
compatible
with
111
and
they
they
won't
work
on
110
and
112
will
just
be
compatible
out
of
the
box.
They
would
have
to
hand
curate
this
stuff
going
forward.
B
Got
more
restrictive
validation,
I
mean
I
mean
I,
haven't
tried,
it
be
honest
in
a
while.
So
maybe
I'm
missing
something
and
I
just
construed,
so
I
guess
it
never
hurts
to
try.
But
every
time
I
tried,
110
I've
come
up
with
okay,
so
okay
and
there
and
to
be
clear
I
mean
I
was
for
Franz
experiment.
This
is
this
is
only
for
the
hosting
cluster.
No.
D
B
Having
member
clusters
different
versions-
and
maybe
the
user
guide
can
sort
of
refer
to
that
like
temporarily
until
we
get
this
resolved,
you
need
to
use.
You
know
a
111
cluster
deployed
via
you
know:
local
cluster,
a
dying
cluster
or
yeah.
What
is
it
GK
or
whatever
your
member
clusters?
There's
not
that
requirement.
C
A
Exactly
that's
pretend,
and
so
what
you
said
listed
Ivan
was
that
we
ideally
should
for
now
remove
the
recommendation
of
mini
cube
in
the
command
set.
Actually
I
have
listed
over
there.
So
we
we
can
do
that,
and
we
can
say
that
it
is,
but
the
current
infrastructure
is
needed
as
1.11
to
install
these
CRE
base
resources,
which,
whatever
restrictions
we
have
yeah
that
can
point
to,
and
we
can
say,
recommendation
that
puppy
for
developers
who
want
to
try
it
out.
A
B
A
B
B
B
A
A
Yeah-
and
you
know
why
I'm
saying
this
is
because
I
did
run,
I
did
do
the
setup
and
then
I
did
run
Danny,
tui
and
I
told
that
there
was
some
time
out
to
choose
and
first
time
couple
tests.
It
failed,
so
I
wanted
to
clean
it
up,
and
it
was
like
pain
that
you
have
to
remove,
which,
after,
if
you
want
to
correctly
do
it,
you
have
to
remove
cluster
or
role
binding
service
account
secret.
All
those
things
and-
and
you
might
see
something-
I
mean.
B
Some
of
the
I
mean
I
did
the
delete
script
I
think
handles
most
of
that,
but
it
only
does
it
for
a
single
cluster.
So
if
you're
doing
multiple
clusters
anyway,
so
I
didn't
yeah
I
can
see
why
you're
more
interested
in
getting
unjoin
working,
because
it
was
kind
of
a
pain
to
fix
it
for
a
single
cluster,
and
then
it
doesn't
solve
multiple.
B
A
A
B
A
Yeah
so
I
remember
that
our
decision
on
having
this
differentiation
was
that
we
will
we'll
start
the
versioning
from
something
like
to
dot,
o
or
B
to
something
like
that
for
this
effect,
so
we
had
support
for
earlier
cube,
fed
until
1.9,
I
guess
with
theaters.
After
that
there
is
no
support.
This
should
have
something
poster
idea.
A
A
B
I'm,
just
I,
don't
know
I
think
I
think
that's
probably
fine.
I
was
kind
of
wondering
as
we
want
to
do
like
cube
that
be
to
just
be
explicit,
but
probably
the
versioning
scheme.
As
long
as
they're
disjoint
like
you'll,
never
see
a
federation
v1
version
greater
than
109
and
you'll
see
v2,
never
less
than
like
111,
then.
Hopefully,
that's
good
enough.
That.
A
B
A
B
A
B
Just
hard
coding
it
for
now,
like
I,
don't
think
we
need
any
fancy
scheme
and
I'm
I'm
not
too
worried
about
the
controller
manager.
Cuz
we're
gonna
have
versioning
in
the
image,
so
I
mean
unless
it's
really
easy
to
just
like
a
lot
of
cube
bill
to
provide
some
sort
of
versioning
mechanism.
I
say
just
do
it
as
simply
as
possible
and
do
like
you
know,
111
alpha-1,
or
something
like
that
as
a
starting
point.
I.
A
B
A
B
I
think
when
the
challenge
is
that
we're
reusing
an
API,
that's
kind
of
tied
to
cube
and
we've
talked
about
this
in
the
past,
so
that
if
by
providing
a
version
that
has
some
sort
of
parallel,
it
gives
sort
of
an
easy
reference
point.
I'm,
not
sure
that's
the
way
to
do
it
long
term,
because
maybe
we'll
want
to
do
something
different.
But
for
now
that
just
seems
simpler.
You
know:
q,
Fed,
v,
111,
alpha
1,
actually,
I
guess
the
Alpha
status
is
maybe
that's.
What's
confusing
like.
D
A
B
D
B
A
B
A
B
I,
honestly,
I,
don't
think
it's
too
critical
to
get
it
exactly
right.
You
just
need
something
to
halfway
sane,
just
to
get
an
alpha
with
the
version.
That's
different
from
existing
versions
of
cube
head.
That's
the
main
thing:
yeah
differentiate
from
cube
fat
in
the
useful
way,
which
I
think
the
cubes
versioning
scheme
is
good.
For
and
however
you
do
that
I,
don't
I
mean
that's
a
suggestion,
but
really
it
doesn't
matter
just
differentiate.