►
From YouTube: Kubernetes Federation WG sync 20180820
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
B
B
Okay,
so
I
think
you
sent
it
on
Sunday,
oh
yeah
right,
they
might
have
not
affected
so
I
saw
the
doodle.
I
saw
the
doodle
before
this,
so
I
think
he's
okay
with
tomorrow's
XP
and
also
know
so.
B
C
B
B
D
B
A
B
A
C
A
C
B
B
E
B
E
B
E
E
B
C
A
So
here's
my
desktop
now,
yes,
okay
great,
so
how
does
the
updated
document
of
pull
modes
based
on
comments
from
from
Maru
from
Khun
Jane
from
your
phone?
So
we
can.
We
can
look
at
it.
So
first
we
can
take
a
look
at
the
background.
So
it's
a
background.
I
have
just
the
add
a
diagram
here
so
from
here.
A
You
can
see
that
basically,
they
want
handles
case
when
the
member
clusters
don't
have
a
inbound
network
from
the
Federation
control
panel,
so
six
mean
that
the
federal
control
panel
we
are
not
able
to
push
data
are
to
the
member
catters.
So
that's
why
I
want
to
make
sure
that
we
we
should
have
the
promote
so
that
remember,
class
I
can
just
pull
some
data
from
from
some
web
host
and
then
it
can
do
some
operations.
So,
since
is
the
background
of
the
this
case,.
H
So
I
think
you
need
to
clarify
that.
Secondly,
if
that
is
the
case,
another
alternative,
if
both
of
the
clusters
are
completely
inaccessible
from
the
internet,
but
have
outbound
internet
connectivity,
they
can
just
connect
at
a
rendezvous
point
being
any
computer
on
the
internet
and
proxy
by
that
rendezvous
point.
So
I'm
not
saying
you
shouldn't.
Do
this
I'm
just
saying
that
if
the
main
reason,
the
only
reason
I've
seen
here
is
that
the
machines
are
not
contactable.
H
A
H
A
H
A
H
C
H
C
Have
a
question:
actually:
oh,
oh
you
all
right
from
technical
point.
It's
you
can
create
some
kind
of
proxy.
You
can
create
a
VPN
or
some
other
solutions,
but
from
security
or
Enterprise
regulations
there
might
be
in
current
connections
are
not
rare
and
not
allowed
at
all.
It's
not
just
you
know
proxy.
H
G
Think
what
you're
saying
is
that
we
can
use
a
rendezvous
point
and
create
a
tunnel
through
it,
where
both
the
cluster,
the
Federation
cluster
and
all
of
the
remote
clusters
connect
via
that
wonderful
endpoint,
meaning
from
their
perspective.
It's
all
outgoing
connections,
not
incoming
connections
and
that
solves
the
problem
learned.
G
H
G
H
Yes
and
that
isn't
actually
stated
as
a
requirement,
so
if
there
is
a
strong
requirement,
then
we
should
they.
There
is
definitely
a
requirement
in
the
Federation
control
plants,
design
that
any
given
cluster
may
be
uncontacted
all
at
any
point,
because
networks
go
down,
clusters
go
down,
etcetera,
that's
the
main
aim
of
Federation
is
to
handle
those
situations
which
is
ever
so
slightly
different
than
a.
H
Something
like
github
solution,
where,
in
fact
there
is
no
requirement
that
they
ever
be
up
at
the
same
time,
but
I
think
the
chances
of
that
are.
You
know
that
you
can
hypothetically
think
of
a
situation
where
you
bring
up
the
Federation
control
plane.
It
makes
a
bunch
of
decisions
it
sends
all
of
that
to
github
somewhere
and
then
shuts
down
for
good
before
the
other
cluster
ever
comes
up
and
then
the
cluster
comes
up,
gets
all
stuff
out
of
github
and
does
its
thing.
A
H
H
If
you
go
to
the
top
of
this
document,
I,
don't
think
that
thing
is
in
the
document
at
the
top
that
the
diagram
at
the
top,
so
I
think
Paul
took
the
die
or
maybe
he
did
end
it.
Okay,
but
anyway,
that
the
point
remains,
it
doesn't
matter.
What's
on
a
document,
the
point
is:
do
we
have
a
requirement
for
it.
A
A
Let
me
let
me
show
the
email
as
well:
I
just
posed
a
question
and
see
mode
cluster
and
yeah
I,
guess
I'm
a
response
from
you
and
also
got
some
comments
from
Paul
and
on
his
own,
some,
this
guy,
who
is
Paul
and
we
we
come
to
the
solution
that
we
want.
You
can
of
pull
mode
and
we
want
to
introduce
github
at
the
said
patio
so
that
we
can
leverage
each
other
as
an
okay
when
you
just
scroll
back
up
to
my
email
at,
don't
remember
exactly
what
I
said
there:
oh
yeah!
H
Are
not
clear
ball?
He
was
killed
here
exactly
so
I'm
saying
it
again
and
if
Paul
wants
to
investigate
it,
which
he
did
and
he's
here
now,
maybe
he
can
explain
what
the
purpose
is.
I
guess
I'm
just
I'm
not
saying
we
can't
build
this
I'm,
just
saying
that
I
don't
understand
why
we're
building
it
and
I
said
that
before
when
I'm
saying
it
again,
yeah.
E
E
E
One
I'm
not
familiar
enough
with
the
specifics
of
the
use
case
around
that,
but,
as
the
strong
one
for
me
is
like
say
that
you
have
say
that
you're
gonna
spin
up
a
new
cluster-
and
you
want
to
you-
want
to
pull
its
information
in
from
some
location
or
treat
clusters
themselves
is
like
replicas
of
one
another
that
that
seems
to
me
like
a
valid
enough
use
case
to
explore
it.
I
I,
don't
have
an
immediate
need
for
this
in
case.
It's
not
perfectly
clear,
though
so,
okay,
so
it's
I
I
share
your.
H
Spirit
pole,
in
that,
if
people
have
reasons
to
build
things,
whether
or
not
we
agree
with
them,
they
should
have
the
mechanism
to
build
them
and
there's
no
question
that
someone
could
build
this
thing.
I,
just
personally
have
don't
haven't
heard
anyone
say
anything
that
leads
me
to
believe
that
there
is
a
need
for
such
a
thing.
Not
that
I
don't
think
there
have
been
anybody
say
anything
to
be
it
and
I'm
trying
to
get
someone
to
say
something.
It
says
we
need
one
of
these
things.
F
Quentin
yeah,
so
the
motivation
for
putting
like
a
configuration
generator
into
the
whole
diagram
in
the
reading
was
that
we
did
have
people
in
federation
v1
saying
I,
don't
really
want
to
give
like
control
over
my
clusters
to
something
that
really
represents
the
central
point
of
failure
and
is
administrative
Lee,
maybe
not
something
that
I'm
willing
to
accept,
and
so
there's
not
like
a
direct
like
there's,
not
nobody
hammering
saying
we
need
this
right
now.
It
was
more
kind
of
like
forward-looking.
F
When
we're
designing
this,
we
want
to
make
sure
that
people
who
said
we
can't
use
Federation
in
its
current
form.
We
give
them
kind
of
in
having
you
to
using
it
if
they
decided
some
things,
they
want
to
do.
I'm,
not
suggesting
that's
a
reason
for
implementing
it
now,
but
at
some
point
in
the
future
there
may
be
a
call
for
yeah.
H
Yeah
I
mean
that
sounds.
That
sounds
like
a
different
reason
and
I
can
understand
that.
So
again,
if
we
just
look
into
that
a
little
more
detailed,
so
so
I
think
what
you're
saying
is
that
you're
replacing
github
as
yours
central
see
replacing
the
Federation
control
plane
as
a
central
point
of
failure
with
github.
Well.
H
F
H
F
F
H
F
H
Think
so
so
I
think
we're
back
again
where
we
started,
which
is.
Do
we
have
a
use
case
that
justifies
a
fairly
significant
number
of
people
reviewing
a
fairly
long
document
and
presumably
some
code
to
come
later
nobody's
gonna,
stop
anyone
from
building
things,
but
but
I'm
personally
resistant
to
spending
large
amounts
of
time
reviewing
things
for
which
we
haven't
identified
a
use
case.
Yeah.
E
And
just
just
to
be
clear,
there
is
no
need
that,
if
someone
wants
to
prototype
whole
reconciliation,
that
that
has
to
happen,
then
in
the
Federation
be
to
repo
like
yeah
I
agree.
One
thing
that
I
would
I
would
very
much
support
is
of
what
we
put
a
type
so
far
determining
what's
what
is
a
function
of
just
the
push
reconciler
and
just
drawing
really
crisp
lines
around
that
functional
area,
so
that
it's
it's
clear
if
you
want
to
write
another
reconciler
or
explore
another.
E
H
H
F
I
would
I
would
concur
I.
Don't
honestly
think
that
the
use
case
described
congratz
be
solved
by
writing
a
bunch
of
software.
It's
Paul
said:
if
someone
once
you
know,
you
guys
want
to
go
build
this.
You
can
do
it
without
putting
into
the
Federation
repo
and
I.
Think
I
share
your
concern,
Quentin.
That
I
mean
given
that
this
is
an
immediate
priority
and
we
do
have
a
lot
of
other
things
for
our
priority
by
devoting
time
and
energy
even
to
reviewing
it
or
do
accepting
something
to
maintain
it
indefinitely.
H
A
Credit
give
you
comment
was
from
some
customers
and
like
without
clarify
here,
because
some
of
the
member
chrisser
don't
have
the
in
bounds
network.
So
we
need
to
make
sure
that
we,
how
they
pull
mode
so
that
the
member
cluster
can
just
pull
some
data
from
from
the
federal
control
panel
from
some
other
places.
A
H
That's
not
really
a
special
cluster,
that's
a
normal
cluster.
The
most
loads
most
clusters
have
have
an
API
point
that
is
accessible,
yeah
right,
bad
it
out
onto
the
internet
via
a
load
balancer.
If
if
they
need
to
get
out
on
or
expose
themselves
on
the
internet,
fire
a
load
balancer
and
that's
what
we
have
ingress
and
services
etc
in
kubernetes
for
the
motors
are
not
accessible
from
the
internet
than
those.
A
Yeah
right,
but
as
from
the
customers
descriptions,
they
don't
want
to
expose
the
API
server
externally,
because
there
was
some
security
issues
so
that
they
only
wanted
to
make
the
classic
I
only
pull
some
data
from
external,
and
they
don't
want
to
make
any
in
bonds.
They
are
observing
because
they're.
So
this
is
a
major
concern
and
a
requirement
from
the
customer.
A
H
A
H
H
A
F
F
C
F
E
In
and
I
think
that
we
really
can
like
we,
we
can
and
maybe
should
do
something
to
clarify
where
the
rest
of
Federation
ends
and
where
push
reconciliation
begins
and
when
I
say
clarify
I,
don't
necessarily
mean
writing
software
I,
don't
I!
Don't
if
I
have
an
accurate
picture
of
everything
in
my
head
I,
don't
necessarily
think
we
would
need
to
do
that,
but
maybe
just
like
a
written
document
for
hey.
If
you
want
to
replace
this
part
of
the
system.
E
This
is
what
you
should
do
like
like
I
said,
it's
I
think
it's
really
important
for
the
overall
long-term
success
of
Federation
that
people
should
be
able
to
use
it
in
the
ways
that
they
want
to
do,
and
so
I
I'm
tending
to
look
at
this
from
more
of
a
standpoint
of
allowing
extension
than
necessarily
you
know
if
I
have
a
use
case
for
X
or
Y
thing
just
so.
My
point
of
view
is
clear:
yeah
I
agree
with
you.
H
You
guys
can
do
whatever
you
like.
I
would
not
pay
someone
to
build
this
software
myself.
I
would
go
and
talk
to
the
customer
and
find
out
why
they
think
proxy
is
less
secure
than
pushing
things
into
github.
I,
don't
know
if
they
propose
to
push
into
github
option
or
not,
but
it's
not
clear
to
me
that
there's
any
good
reason
to
push
it
into
get
up
there.
There
are
other
reasons
to
be
clear
that
yeah
such
a
thing
might
make
sense.
H
H
E
E
Workloads
and
resources
in
a
cluster
and
I
would
imagine
that
there
are
people
that
are
already
using
Cuba
plier,
that
if
Federation
can
write
to
something
that
Cuba
player
can
use
would
more
easily
integrate
Federation
into
their
into
their
community
of
practice
around
kubernetes.
But
that's
that's
mostly
me.
Just
kind
of
speculating
so
yeah
I
want
one
thing
that
I'm
sensing
here
just
to
share
this.
Is
that
a
lot
of
times
when
a
lot
of
times
it
can
be
very
possible
to
have
a
use
case.
E
That
seems
like
perfectly
evident
within
your
group
or
your
organization,
but
that
you
have
trouble
articulating
to
folks
that
are
outside
of
that
context
and
I
think
I
might
be
happening
here
and
I.
Think
that
is
one
of
the
things
that
is
sort
of
pregnant
in
this
conversation,
so
I'd
be
happy
to
have
another
chat
a
little
bit
more
focused
on
like
the
requirements
ends
of
this.
E
If
people
are
interested
in
doing
that
and
offer
myself
as
like
a
sounding
board
for
like
exploring
the
motivations
a
little
bit
more
and
seeing
if
there's
a
way
that,
like
we're,
perhaps
talking
past
one
another-
and
there
there's
a
use
case
here-
that
people
would
be
more
broadly
interested
in
if
it
was
articulated
in
kind
of
a
first-class
way.
I'd
be
happy
to
do
that.
E
E
So
I'm
only
saying
I'd
be
happy
to
have
like
a
more
detailed
conversation
that
focuses
more
on
the
requirements
ends
of
it.
Then,
like
the
Federation
technology
aspect
of
it-
and
you
know
a
lot
of
times
having
that
kind
of
outside
perspective,
can
be
really
helpful
for
finding
the
pieces
of
information
that
make
the
difference
between
you
know.
Okay,
you
can
do
this
or
Wow
I
actually
want
the
same
thing
as
you
do.
Does
that
make
sense,
yeah.
E
So
we're
at
about
we're
at
a
little
bit
over
half
time
now,
I
wonder
if
people
want
to
use
the
remaining
time
in
this
meeting
to
talk
about
the
query
API,
because
I
believe
that
was
also
on
the
agenda
for
today
yeah
and
it
seems
like
that's
an
area
where
we
probably
have
broader
agreement
on
the
motivation,
so
that
that
might
be
a
productive
thing
to
use
time
on.
In
the
remaining
time
we
have
today.
A
E
That's
modular
modularized,
to
the
extent
that
if
you
want
to
replace
a
piece
of
it
that
you
can
do
that
easily
and
so
I
I,
don't
I
I,
don't
think
anybody
is
saying,
don't
don't
explore
this
idea,
I
think
what
I'm
hearing
from
other
people
on
the
call
is
that
they're
not
convinced
that
it's
a
use
case
that
they
feel
like
in
your
place.
They
would
want
to
develop
software
for
and
so
I
think
the
takeaway
from
this
part
of
the
discussion
is
probably
two
part.
E
But
you
know
again:
I
I,
don't
think
that
we've
really
captured,
probably
the
full
the
full
list
of
important
aspects
about
the
use
case,
because
I'm
hearing
I'm
just
hearing
a
little
bit
of
a
disconnect
on
that,
and
so
the
second
point
I
would
say,
is
that
I'm
very
happy
to
offer?
You
know
myself
is
like
a
sounding
board
or
just
an
outside
perspective
if
we
maybe
want
to
have
like
a
smaller
conversation,
that's
more
focused
on
the
use
case
here.
E
I
am
this
is
this:
is
this
is
kind
of
what
I
was
referring
to
by
having
some
guidance
for
people
that
want
to
extend
or
replace
parts
of
Federation
I
think
it
would
probably
be
really
productive
to
I'm,
not
in
an
area
with
a
white
board
today,
but
this
seems
like
if
we
can
add
a
white
board
to
the
discussion
as
as
a
tool
maybe
on
on
both
ends.
So
we
can
draw
some.
E
A
E
A
C
C
C
A
C
Know
it's
you
sharing
it,
so
you
can
continue
it's
okay,
but
please
start
form
options,
free
action,
it's
the
streets.
You
know
the
idea
of
which
query
API,
it's
a
even
in
the
iteration
documentation
for
to
provide
some
examples,
but
you
use
a
key
to
use
scripts
customer
and
say
based
on
them.
You
configure
information
without
any
information
about
cluster
cluster
itself.
For
example,
the
status
of
the
clusters
get
to
has
internal
cluster
registry,
but
user
don't
have
access
to
the
is
not
able
to
get
informations,
in
addition,
say
scripts.
C
Actually
the
idea
to
create
this
abuse
at
the
first
option.
It's
to
change
in
user
experience
to
provide
the
give
is
that
user
can
create
declarative
information,
what
he
wants
to
get,
for
example,
and
get
nodes
with
specific
conditions
from
all
classes,
and
this
information
can
be
really
clear
up
dated.
C
C
A
A
C
C
Let's
provide
the
ability
to
to
customer
create
some
kind
of
a
Buick
to
create
the
view,
so
it
can
be
read
queries
with
information
about
information
from
each
cluster.
The
information
should
be
provide.
For
example,
we
can
provide
cluster
selector.
This
much
plasma
image
cluster
means,
but
maybe
not
all
clusters
are
relevant
for
that
specifically,
when
we
can
provide
information
about
the
conditions
from
the
cluster
registries
upon
the
clusters
that
are
self
conditions
will
be
well
and,
in
addition,
here
it's
a
conditions
that
they
provide
some
kind
of
beautiful
eight
months.
C
It
can
be
one
man,
one
shot
just
like
a
comment
from
the
script,
get
information
or
it
can
be
periodically
these
periodical
updates.
Another
options
can
be
some
static.
Like
a
bank
based
this
water
price
information
will
be
synchronized
and
the
gate
seeing
the
kind
of
information.
What
kind
of
information
should
be
provided
like
reports
notes
or
something
else,
so
it's
easy
I
say
he's
like
complimentary
to
the
Federation
abilities
currently
Federation's.
C
F
C
C
F
F
Mean
I
we
talked
about
in
this
nopales
and
it's
like
there's
all
kinds
of
different
ways
to
provide
visibility
into
Federation
members.
This
is
one
possible
way
and
to
me
the
fact
that
I
don't
have
a
clear
idea
as
to
why
this
is
justified
suggests
that
there's
information
missing
on
the
user
side
of
things
am
I
the
only
one
that
feels
that
way
is
there.
Anybody
else,
yeah.
H
I
had
a
similar
comment:
I
looked
at
the
document
and
the
use
cases
of
basically
as
far
as
I
can
tell
gke
features
and
I'd
I
stand
corrected,
but
I
believe
that
GK
he
did
this.
Just
in
the
user
interface
I
haven't
actually
I'd
look
at
their
API,
but
I
I
think
that
all
they
did
was
they
used
the
existing
kubernetes
api
to
build
a
UI
across
multiple
clusters,
so
yeah
similarly
I
happen
to
think
there
are
good
use
cases
for
a
federated,
read
API,
but
I
agree
that
they're
not
expressed
in
the
document.
C
H
C
F
C
F
There
are
concerns
around
just
using
those
credentials,
raw
against
all
the
member
clubs
like
this
was
discussed.
As
for
the
document,
I
believe
you
had
access
to
around
mic.
Okay,
I
want
to
access
these
clusters
whose
credentials
do
I
use
it.
Do
I
use
the
credentials
of
the
user
who's.
Invoking
the
query
do
I
use,
you
know:
do
I
impersonate
them
do
I
give
like
do.
F
I
give
just
kind
of
global
access
to
everybody
like
there
were
all
kinds
of
concerns
around
authentication,
so
I,
don't
think
the
answer
is
as
simple
as
I'm
just
going
to
reuse
exactly
what
Federation
is
using
to
interact
with
the
questions,
nothing
you
couldn't
but
I.
There
seem
to
be
a
lot
of
concerns
from
the
siga
group
about
that
approach.
Yeah
I,
think
that's
why
I'm
raising
that?
Like
it,
you
know.
Yes,
you
want
to
use.
F
H
Now
I
think
I
understand
what
what
you
were
saying
earlier.
You
were
saying
that
you
cannot
get
the
status
out
of
the
cluster
registry
because
it
doesn't
know
what
the
status
is
all
right.
Yes
and
the
reason
it
doesn't
know
is
because
it
doesn't
have
credentials
to
access
the
cluster
and
and
find
out
what
its
status
is.
Okay,
now
I
understand,
and
so
you
want
to
put
that
status
somewhere
else
by
some
other
mechanism.
H
H
F
You
talk
about
cluster
conditions,
that's
actually
a
function
of
the
federated
clusters,
so
the
cluster
registry
cluster
is
simply
a
list
of
API
endpoints
effectively
and
then
the
federated
cluster
is
what
Federation
is
using
and
so
there's
an
active
controller
that
updates
the
federated
cluster
with
cluster
conditions.
So,
if
you're
using,
if
you
wanted
to
create
some
sort
of
mechanism
to
query
that
I
mean
you
just
have
access
to
the
federated
clusters-
and
you
can
achieve
what
you're
talking
about
here
but
I'm
not
saying
that's
your
only
purpose
for
having
read
access.
F
Absolutely
I
mean
the
server-side
thing
might
have
utility,
but
it
also
sort
of
it
requires
you
to
duplicate
like
you
have
to
have
a
comprehensive
mechanism
for
querying
versus
oh
you're,
just
scripting
you
Ketel
commands.
Assuming
you
have
access
to
the
underlying
clusters.
I
mean
there
is
kind
of
a
general
problem
around
a
sonication
for
multiple
clusters
and
I.
Think
if
you
were
really
serious
about
maintaining
like
a
lot
of
clusters,
the
single
sign-on
solution
would
probably
be
a
good
idea,
and
then
you
know
you
can
grant
permissions
to
some
sort
of
read.
H
Yeah
I
mean
just
to
be
clear.
This
applies
to
Federation
and,
in
fact
kubernetes
as
well.
There's
nothing
that
kubernetes
does
that
you
can't
do
with
a
bash
script.
The
question
is
just
how
complicated
your
bash
script
gets
and
what
permissions
your
bash
script
needs
in
order
to
get
its
job
done,
and
so,
from
one
point
of
view,
kubernetes
is
really
just
a
way
of
making
your
bash
script
simpler
and
taking
away
some
of
the
credentials
that
you
would
need
in
order
to
be
able
to
get
the
job
done
and
putting
them
somewhere
else.
H
C
H
H
F
Is
true,
it
just
occurred
to
me
that
I
think
the
way
that
we
run
when
you
start
the
controller,
men
or
I
think
it
starts
the
cluster
controller,
no
matter
what
we
don't
even
offer
the
option
to
disable
it.
So
if
you
were
to
run
the
Federation
controller
manager,
disabled,
you
know
the
question
Seiler
scheduling
in
DNS,
you
would
still
get
cluster
conditions
for
giant
clusters.
Yeah.
A
H
So
I
understand
that
part
I
just
want
to
make
it
clear
to
you.
If
your
end
goal
is
the
things
specified
in
the
use
cases
at
the
top
yeah
you're,
all
user
interfaces
and
and
which
are
all
provided
today
by
gke
I,
believe
none
of
that
required
any
of
the
kind
of
stuff
in
this
design
that
that
is
just
a
user
interface
on
top
of
the
existing
communities.
H
A
D
F
There's
no
suggestion
in
the
GK
control
panel
of
the
implementation
I
think
it's
very
unlikely
that
they've
implemented
a
kubernetes
api
to
access
the
status
of
other
underlying
clusters.
It's
far
more
likely
they're,
actually
calling
all
the
registered
clusters.
So
to
me
this,
you
know
we
want
some
feature
to
get
access
to
this,
but
it
doesn't
suggest
the
implementation.
H
Yeah
I,
agree
and-
and
one
other
comment
was
just
before
we
go
into
the
detail,
so
so
we
struggled
with
this.
We
did
a
lot
of
this
design
earlier
and
and
a
lot
of
the
there
are
some
fairly
difficult
technical
issues
to
solve,
like
latency
and
asynchronous
behavior,
and
caching
and
permissions
as
Larry
mentioned.
H
If,
however,
your
Federation
controllers,
which
need
to
have
snow
which
pods
or
in
which
clusters
to
be
able
to
make
decisions,
if
they
need
access
to
that
and
they
are
perhaps
polling
or
watching,
that
information
that
sort
of
suggests
a
very
different
implementation,
potentially
so
I
would
encourage
you
before.
You
start
designing
anything
to
just
have
a
very
clear
understanding
of
what
you,
what
your
actual
use
cases
are
and
and
which
ones
you
are
not
supporting,
because
right
now,
the
ones
you
have
specified
turn
motivate
for
doing
anything.
H
H
H
A
H
Of
the
user
requirements
or
yeah
make
any
difference.
The
the
difference
between
v1
and
v2
is
more
about
implementation.
The
use
case
right
yeah
right.
Yes,
so
the
question
is
well
so
off
the
top
of
my
head.
I,
don't
remember
all
of
them,
but
other
people
did
lots
of
work
and
I.
Don't
have
the
documents
in
my
head
right
now,
but
but
one
use
cases
is
controllers
that
need
that
information
from
the
clusters
who
may
need
it,
for
example,
how
many
nodes
are
there?
How
big
are
they?
Is
this
thing
in
a
schedule,
etc?
H
H
So
do
you
have
a
piece
of
software,
for
example
that
runs
on
top
of
kubernetes,
that
that
does
something
like
list
nodes
or
pods
to
do
something
with
them
and
it,
and
it
can't
run
across
multiple
clusters
today,
because
it
we
don't
have
this
axis
5
Federation.
If
you
have
such
a
such
a
piece
of
software,
that
would
be
useful
to
understand
what
it
needs.
H
Just
to
be
clear,
you
don't
have
to
do
anything.
You
can
go
and
build
this
stuff
tomorrow.
If
you
like
yeah
we're
just
trying
to
advise
you,
you
know
what
what
you
probably
want
to
build
is
determined
by
what
you
want
to
use
it
for,
and
it's
not
clear
what
you
want
to
use
any
of
the
stuff
for
so
we
can't
really
suggest
whether
or
how
you
should
build
it
until
we
understand
that
I
guess,
that's,
maybe
the
best
way
of
putting
it.