►
From YouTube: Multi-Network community sync for 20230322
Description
Multi-Network community sync for 20230322
A
B
A
B
A
Was
a
lot
of
comments
coming
from
Dan
Winship
I
addressed
all
of
them
updated
the
cap
so
that
it's
better
reflected
and
he
had
very
good
pointers
on
what
we
wanted
to
do
and
how
the
kind
of
kind
of
form
of
the
cap
should
look
like.
So
there
was
a
few
tweaks
I
did
here
and
there
I
am
trying
to
push
from
my
side
on
team
to
review
this
as
well,
and
my
goal
is
to
get
it
kind
of
merged.
A
I
would
appreciate
anyone
else
from
this
group
as
well
review
it
again.
If
there
is
some
things,
that's
not
okay,
but
currently
the
the
we
are
in
the
just
waiting
game
for
someone
to
sit
down
and
review
it.
I
haven't
pushed
it
like
for
the
last
few
weeks
because
there
was
a
kubernetes
release.
So
currently,
most
of
the
folks
were
reviewers
were
busy
with
the
cap,
reviewer
reviews
and
now
the
change
reviews
for
the
next
release
of
kubernetes.
C
A
Yes,
so
the
requirements
Doc
is
out
there.
It's
closed
for
reviews
and
I
can
see
in
the
in
the
red
I'm
saying
to
go
and
go
to
PR
and
and
comment
on
PR
right
now.
The
all
the
development
and
comments
right
now
is:
don't
group
the
pr
for
the
requirements
cap.
So
please
please
look
at
that
and
give
it
a
read
if
you
can
to
yeah.
Look
at
it
again,
if
there's
some
other
gaps.
Let
me
know.
D
And
then,
let's
sorry,
one
one
question
about
process,
so
we're
talking
about
API,
it's
been
very
much
about
pod
networking
about
how
the
Pod
is
going
to
change.
I
realized
reading
through
it
that
I
haven't
seen
anything
about
how
any
of
the
other
resources
changed.
Is
the
plan
we're
going
to
get
pod
Network
worked
out
and
then
go
back
and
fill
in
some
of
the
other
resources
and
how
they'll
change.
A
Look
at
the
requirements
cap,
my
proposal
there
is
to
and
something
that
you
saw
last
week.
I
was
last
two
weeks
ago.
I
was
showing
the
roadmap.
We
are
phasing
out
the
whole
thing.
So
our
phase
one
phase,
one-
is
all
about
just
defining
the
API
and
being
able
to
reference
that
in
the
pod,
and
that's
it.
Let's
figure
that
one
out
later
on
we're
gonna
look
at
other
resources.
What
do
you
have
in
mind?
Maybe
I
will
start
that.
D
Yeah
yeah
so
I'm
reviewing
this
I,
don't
see
any
issues
yet
I'm
just
thinking.
Oh
the
reason
that
I'm
interested
in
this
is
that
we've
got
some
some
slightly
weird
use
cases.
I
haven't
even
got
teams
to
probably
write
down
for
me
in
full
detail
from
from
our
side
about
things
like
sriov
and
Numa
attachments
and
so
on,
and
how
is
that
all
going
to
be
possible
to
do
this
through
what
we're
defining
now
so
I'm
trying
to
build
in
my
head
a
picture
of?
D
Are
we
going
to
have
the
right
information
available
to
do
all
those
things
now,
I
suspect
the
reality
is
we
will
I
mean
there's
two
ways
you
could
imagine
handling
that
one
is
to
do
some
prototyping
now
and
I.
Think
that
might
be
a
sensible
thing
to
do
in
parallel
for
us
to
do
the
other
way
to
handle,
it
is
to
say:
well,
we
will
put
down
what
we
think
we
need
and
then,
as
later
on,
we
might
need
further
information
or
further
capabilities.
We
might
extend
things
so.
A
Again,
I
think
and
just
to
be
more
clear.
The
requirements
cap
as
well
includes
all
the
use
cases
and
all
the
requirements
that
we
want
to
handle
through
this
effort
so
and
I
want
to
get
it
merged
so
that
we
don't
get
distracted.
I
know
folks
can
come
in
and
at
the
various
times
of
points
of
this
effort
and
I,
don't
want
to
get
distracted
by
okay.
Now
did
we
talked
about
this
or
that
and
then
I
would
say,
no
stop.
The
discussion
go
to
this
Doc,
and
this
is
what
we
are
covering
all
right.
A
So
we
are
still
we
are
still
in.
You
are
selling
good
point
where
you
can
add
your
comments
into
that
cap
and
when
it's
merged,
that's
it
I'm
I'm,
not,
and
that
will
be
my
always
response
to
any
try
attempts
to
did
we
thought
about
those
use
cases.
I
would
say
no
that
something
is
already
defined.
This
is
the
cap,
and
this
is
what
we
are
focusing
only
on.
A
A
D
The
the
other
example
I
can
think
of
where,
like
I
said,
I
don't
have
a
a
hard,
a
hard
requirement
here
and
I
think
is
when,
when
we
add
things
to
pod,
it
is
not
just
what
networks
they're
in,
but
potentially
what
parameters
they
need
to
be
provided
and
so
on
and
I'm
trying
to
get
my
head
around
whether
we've
got
enough
flexibility
in
here
so
I
sure
I
guess
we're
at
the
right
stage.
For
that
now,.
A
Yeah
I
think
last
we're
gonna
discuss.
You
will
see.
I
have
an
example
below
that's.
Gonna,
probably
show
you
what
what
I
have
in
mind.
So,
let's
maybe
go
through
the
finalize.
The
spec
and
I
I
have
some
example
over
there
to
kind
of.
Maybe
I
can
just
show
you
that
so
that
you
can
I
create
an
example
here.
What
how
I
would
imagine
this
would
look
like
and
sorry
folks
I
borrowed
motus
as
an
example.
I'm
not
sure.
Is
that
okay
or
not
hope
now
hope
it
is
okay.
A
So
what
I
would
imagine
is
something
like
that,
where
the
network
object,
where
the
very
basic
what
I
want
to
do,
and
in
this
case
I'm
assuming
I'm
gonna
use
external,
because
in
my
network
attachment
in
my
my
not
I'm,
gonna
use
whereabouts,
so
I
don't
care
about
any
of
the
kubernetes
attachment
of
ipam
I'm
gonna
do
externally
down
ipam
done
by
whereabouts.
This
is
just
pointer
to
the
object,
and
here
my
provider
multi,
saying
that.
A
Okay,
this
this
object,
this
whole
object
is
handled
by
maltus
and
nobody
else
should
touch
it.
That's
how
and
that
doesn't
have
to
be
like
this.
This
is
just
my
example.
Probably
the
multus
effort
is
going
to
figure
out
something
correct,
but
that's
how
basically
an
example
of
how
that
would
look
like
is
Tomo.
B
B
A
So
a
cluster
object
in
reference,
namespace
objects,
quite
fine,
just
a
matter
of
your
controllers
that
will
handle
this
object,
having
the
full
permissions
to
either
look
at
namespace
object
and
look
at
the
other
name
spaces
right.
So
that's
quite
kind
of
straightforward
and
there
shouldn't
be
any
issues
right.
A
You
point
out
to
the
specific
to
a
specific,
just
namespace
and
just
read
that
right
and
then
you
handle
it
accordingly
in
your
controller
so
and
that's
a
good
point
Tomo
and
that's
a
good
example
even
better,
because
now
yes
having
a
cluster-wide
object
referenced
here,
it's
easy
how
about
namespace
here
we
can
do
that
as
well,
and
this
is
all
the
parameters
forever
is
a
standardized
field
so
that
have
all
those
options.
So
that's
how
this
would
look
like.
A
A
A
All
right,
I
see
my
comment
here
on
the
provider.
I
haven't
done
this
one
I
will
write
it
down,
but
from
our
last
discussion
just
to
quickly
remind
that
a
provider
is
just
it's,
it's
a
signal
for
the
implementation
side
and
it
Def,
and
it
says
whether
I
am
to
use
this
object
or
not,
and
it's
up
to
the
as
well
implementation
of
how
they
behave
when
this
object
is
not
provided
right.
A
Do
I
am
I,
am
I
a
kind
of
eager
in
this
case
or
and
just
read
everything
if
it's
not
if
it
specifies
for
me
and
when
it's
not
present
or
do
I
require
provider
to
be
always
specified.
So
that's
something
that
implementation
has
to
identify,
whether
they
how
they
treat
this
and
respect
this.
This
provided
field.
E
A
Have
ipam
so
what
about
so
today?
We
don't
have
that
option
right
today
in
kubernetes,
we
don't
have
option
to
separate
the
two
and
to
identify
what
I
want
to
do
for
for
for
specific
family
and
I
I
can
have
I
would
imagine
I
have
a
case
where
okay
V4
just
do
with
standard
when
we
are
standard
cases,
but
for
V6.
Do
me
a
slack
or
something,
and
do
it
through
external
means,
so
I
want
to
have
the
flexibility
to
distinguish
what
I
want
to
do
for
for
a
specific
family.
It.
A
Exactly
so
and
and
clamping
that
together,
that's
I'm
telling
you
that's
I,
see
from
Ours
from
my
my
my
our
internal
use
cases
that
this
is.
This
is
something
that
we
would
want
to
have
as
well,
where
okay
I
want
to
kind
of
distinguish
them,
and
so
we
don't
have
that
that
capability
today
on
the
KCM
today,
when
we
do
that
in
in
by
the
by
just
the
default
for
the
default
Network
right,
we
don't
have
the
capability
of
splitting
and
it's
not
very
clear
on
what
is.
E
D
A
Other
aspect
to
this
is
now
I
specify
what
type
is
I
want.
You
see,
I
I
can
and
I
can
have
a
explicit
identification
of
okay.
I
want
ip4
I
want
dual
stack,
so
I
specify
both
or
I
want
type
ip6.
Only
so
my
network
has
a
has
a
direct
identification
of
what
IP
is
IP
families
I
care
about
all
right
on
Spotify.
Neither
then
I
don't
care
about
any
IPS
I
just
do
L2
or
I
just
do
PCI
through
a
PCI
password
or
something
and
I
don't
care
about
IPS.
E
E
A
I
think
the
reasoning
behind
like
being
able
to
identify
what,
which
case
I
want
out
of
the
four
right,
because
I
can
have
specific
version
of
the
family
or
dual
stack
or
not
right.
So
now,
I
have
a
capability
to
explicitly
specify
that
right,
explicitly,
say:
I
want
duals
like
I,
want
V4
I
want
V6
or
you
know,
I,
don't
want
anything
the
other
option
for
that
would
be
okay,
let's
collapse
those
and
have
another
flag
stating
I,
know
IP
family
or
something
like
that
right.
A
Where
I
can
explicitly
say:
okay,
I
won
V4
I
want
V6
I
want
none!
Something
like
that.
There
could
be
another
field
for
that
right,
but
I
think
having
those
two
since
those
serve
as
the
kind
of
capability
of
ipam,
what
sort
of
iPhone
they
can
as
well
specify
and
serve
us.
What
sort
of
families
I
care
right
in
this
in
this
network.
E
A
I
expect
both
IPS
from
this
from
the
cni
or
I
expect
just
one
IP
or
none
that's
what
this
will
drive
towards
right,
because
otherwise,
if
your
implementation
doesn't
provide,
those
I
should
fail.
That's
how
I'd
imagine
later
on
that's
kind
of
how
would
this
behave
where
okay,
you
specifies
those
this
network,
so
I
expect
for
this
interface.
I
expect
this
many
IPS,
regardless.
A
Will
not
cover
the
case
where
you
say
multiple
episodes
right
now:
I've,
let's
talk
about
just
single
IP
and
let's
I
I
wouldn't
want
to
go
and
discuss
that
case.
For
now,
this
I
would
assume
at
least
one
IP
right
and
probably
when
we're
gonna
talk
about
status,
about
how
your
report
status-
and
this
is
where
we're
going
to
talk
about
pause
and
and
now
the
statue
says.
This
is
where
we
probably
have
to
into
get
into
that
discussion,
but
yeah,
maybe
I
would
I
want.
B
Yeah
one
question
yeah
regarding
this
six,
so
they
that
I
have
the
two
stuff.
The
one
is
the
clarification.
So
if
the
usual
case
at
that
time,
the
user
should
specifying
the
boss,
why
not
and
then
next
sex?
The
question
is
about
the
what
the
the
priests
the,
what
how
they
say,
maybe
that
this
is
just
comment.
Please
add
via
comment
via
what
the
optional
means.
This
is
the
just
the
left,
the
default
or
just
the
disable
I
mean
the
type
of
four
is
they
are
not
added.
B
C
So
I
think
I
mean
since
and
is
not
here,
but
I
think
we're
mixing
together.
The
sort
of
lay
two
properties
under
A3
purposes
of
the
network
I
mean
on
a
lead.
Two
we
look
at
V6
I
can
have
as
many
layer
3
networks
as
I
want
or
I
can
just
sort
of
have
a
normal
ethernet
I
think
that's
what
Ed
called
it,
because
it's
ethernet
and
IP
right.
C
This
is
really
about
the
service
you
want
to
offer,
and
I
would
say
that
if
I
offer
a
Visa,
if
I
offer
an
IP
service,
then
I
can
sort
of
describe
it
so
say:
oh
there
is
one
IP
network
on
there.
Maybe
I
can
then
say:
I
want
server
line
layer,
I
mean
IP
Networks
and
it's
really
Upton
to
the
implementer.
If
you
decide
special
V6
to
run
them
over
the
same
physical
L2
or
not
so
I
think
there's
a
little
assumption
that
this
is
a
to
me.
A
C
F
A
Your
and
how
your
pod
is
going
to
handle
that,
if
you
are
have
a
demon
set
to
some
agent
that
okay
I
say
that
this
pod
will
have
those
five
networks
and
then
I.
Just
when
I
when
I
kick
it
off,
I
don't
have
to
create
multiple
inter
physical
interfaces
inside
the
Pod,
because,
like
the
most
implementation
imagines
that
and
does
that
right
will
create
separate
interface
pipeline
into
that.
But
in
your
implementation,
that
can
mean
I
will
create
multiple
IPS
on
that
single
interface.
C
C
C
A
C
A
F
C
What
I'm
scared
of
sort
of
is
that
what
I
said
before?
If
I
want
to
have
an
Ethernet
how
to
solid
L2,
then,
then
it
needs
to
be
a
flat.
Network
right.
I
need
a
broadcast
domain,
basically
preferred
IP
network.
It
can
be,
it
can
be
delivered
through
a
hierarchy
of
networks
and
routing
and
so
on
and
since
I
wanted
IP
I,
don't
really
care
about
what
the
underlying
topology
is.
It
can
be
a
flat
L2
that
is
delivered
over
or
it
can
be
a
a
routing
tree
or
something
like
that.
C
A
A
I
would
imagine
it
will
be
done
through
through
the
parameters.
Reference
I,
don't
think
we
should,
because
what
you're
asking
now
to
kind
of
give
or
identify
whether
it's
an
ll203
and
I
say
no,
because
we
we
Define
that
that
kind
of
parameters
I
can
have
an
implementation
that
only
does
L2
and
I
don't
care
about
the
other
one
or
you're
going
to
have
implementation.
That's
going
to
have
both.
So
what
I'm
saying
is
I'm
I'm
offloading
this
to
this
parameters.
Reference.
C
C
A
C
I
think
this
level
I
think
if
I
don't
feel
it.
If
there's
nothing
there,
I
have
to
trust.
There
is
an
Ethernet,
but
still
the
problem
is
sort
of.
If
someone
has
so.
The
thing
is
this:
if
there
is
a
network
and
someone
has
filtered
in
the
V
for
a
V6
iPad
there
that
there
are,
how
do
if
I
just
want
to
use
an
Ethernet
Network?
How
do
I
know
if
I
can
use
this
network
or
not
I?
Don't
because
I
don't
know
so
so
I
don't
know.
If
this
network
can.
A
C
A
C
D
D
A
C
F
C
F
C
E
C
A
All
right,
let's
going
to
the
next
field
in
the
last
one
in
the
in
the
object-
and
this
was
just
straightforward
copied
from
maltus-
and
this
is
all
about
just
stating
some
sort
of
giving
some
sort
of
access
permissions.
Maybe
this
is
something
that
not
how
it
should
be
done.
We
we
in
our
in
our
requirements.
A
B
A
C
We
need
so
sorry.
I
was
new.
We
need
to
bring
in
the
discussion
around
what
happens,
because
this
is
only.
This
is
only
checked
right
when,
when
you
do
the
action,
so
let's
say
that
you
would
you
would
change
the
allow
the
rules
after
something
has
been
set
up.
We
also
need
to
point
that
this
needs
to
be
checked
either
in
the
cubeless
cryo
or
by
hcmi.
We'll
have
a
discussion
already
at
the.
D
C
F
C
C
B
A
Anyone
has
other
comments
like
General
comments
or
anyone
that
sees
caps.
Are
we
missing
something
here?
I
will?
Let's
not,
maybe
if
anything
please
link
is
in
the
dock.
If
you
have
some
other
comments,
leave
them
and
we
can
talk
through
comments.
If
you
see
unless
you
know
now,
then,
and
then
let
us
know,
if
not
then
just
reread
this
again
and
look
at
this
object
and.
C
See
if
there's
something
specific,
you
took
away
the
routes
right,
but
didn't
we
discuss
to
have
not
the
actual
routes
but
which
route,
which
routes
this
network
object
would
sort
of
which
which
no
not
throughout,
which
address
Scopes.
That
can
be
route
that
will
be
routed
through
this
network
or
have
routes
for
I
mean
that
you
would
be
able
to
see
that
okay,
this.
A
A
Yeah-
and
it
is
so
not
not
exactly
because
iPhone
today
is-
is
handled
for
the
default
Network.
We
have
ipam
Handle
by
the
default
for
the
default
Network.
So
this
is
I'm
copying
only
that
piece
if
I'm
doing
it
or
or
someone
else
is
doing
it,
I'm
not
saying
how
it's
done,
I'm
not
saying
what
cider
it
is,
I'm,
just
saying
how
who
is
doing
it?
A
That's
the
the
the
the
the
the
single
only
information
I'm,
saying,
because
what
I'm
trying
to
say
is
and
what
I'm
trying
to
get
as
at
all
as
well,
keep
in
mind
that
this
object
eventually
should
represent
the
existing
default
Network.
This
object
has
to
do
that
right.
So,
whatever
we
have
there,
it
should
represent
that
part,
and
that's
that's
something
that
this
that
we
have
to
have
right
so
KCM.
Today
we
have
a
KCM
that
in
the
arguments
in
the
KCM
you
can
kubernetes
cluster
manage
cluster
control
manager.
A
You
specify
whether
you
want
to
have
the
iPhone
controller
enabled
or
not
for
that
Network
and
that's
of
course,
Global
for
just
this
default
network,
but
it's
there
and
basically
this
is
what
exactly
this
translates
to.
This
is
the
same
thing:
okay,
with
additional
caveat
where
and
the
caveat
being
a
I
can
specifically
say
what
5p
families
I
care
and
there's
another
option
which
wasn't
it's
not
present
in
KCM
and
that's
which
that
I
don't
have
to
specify
anything
right.
A
A
That
are
common
here,
so
you
can't
remember
that,
okay,
so
for
that
point,
that's
I
think
the
birming
so
default,
Network,
probably
depending
on
who's
implementing
fit.
It
might
not
have
even
these
parameters
ref,
but
it
might
I'm
I'm,
not
sure.
Probably
it
can
and
then
a
provider-
that's
that's
just
optional,
probably
for
the
default
Network.
It
should
be,
or
maybe
now
it
it
has
to
be
specified
of
something
yep.
That's
I,
envisioned!
A
A
The
default
Network
all
right,
the
parameters,
reference
I,
think
this
is
kind
of
straightforward
or
are
there
any
other
comments
to
this
I
think
this
is
just
describing
a
standard.
I
think
this
is
standard
I
copied
from
some
thinking
to
API
or
something.
This
is
a
reference
to
another
object
in
kubernetes,
so
I
think
it
handles
all
the
possible
cases
all
right,
and
the
last
thing
is
the
status
of
the
network,
as
you
can
probably
saw
in
my
example.
A
What
I
want
to
do
is
do
introduce
conditions
into
this
object
and
through
conditions.
This
is
the
standard
conditions
that
we
have
in
pods
and
every
other
kubernetes
objects.
I
want
to
introduce
you
to
types
ready
and
another.
One
is
params
ready
and
those
conditions
with
with
identify
API
side
of
Readiness.
Keep
that
in
mind.
This
is
only
for
the
API
side
of
Readiness,
where
all
the
parameters
are
present.
A
They
are
correct
for
my
implementation,
so
I
have
a
provider
defined
here
provider
X
has
to
look
at
the
object
and
say
yep
everything's
fine.
With
this
object,
no
misspelling,
no
nothing
I
would
set
the
ready
flag,
and
this
is
only
for
that.
This
is
not
for
a
pod
to
say
apart.
It
will
eventually,
but
it's
not
to
say
that,
oh
yes,
this
network
is
available
on
all
the
nodes.
No,
this
ready
is
because
it's
a
centralized
point.
We
cannot
have
a
pair
node
information
about
status
of
that
Network
on
the
subject.
A
It
it
kind
of
is
a
bit
as
well,
because
you
know
if
I
I
misspell
something
here
or
I'm
missing
some
parameters.
It
will
say
the
network
is
ready
and
your
implementation
should
appropriately
react
to
say.
Okay,
this
network
is
not
available
on
my
notes,
because
the
curve
core
object
is
not
ready
for
whatever
reason
and
that
reason
might
be
You
misspelled
something.
But
it's
still
you
shouldn't
try
to
implement
at
this
network.
All
handle
this
network,
because
there
is
some
mistake
in
the
in
the
API
right
so.
D
C
Let's
go
with
your
thoughts,
so
the
the
Reddit
there
means
that
you
can
communicate
with
API
server
about
this
network
right
I
mean,
of
course
you
have
asked
for
status
before
so.
You
know
something
about
this.
You
can
go
go
to
the
next
thing
and
that
there's
no
sort
of
State
about
the
actual
Network
behind
I
mean
metric
routes
before
but
I
mean
basically
up
and
down.
That
would
be
somewhere
else
and
I.
Think
that's.
A
A
C
A
C
The
the
owner
of
the
network
has
another
way
to
to
bring
down
the
whole
network,
yeah
yeah
the
operator
part
of
it,
but
so
that
means
that
you
shouldn't
be
able
to
change
the
admin.
States
of
the
network
should
be
able
to
report
it
somewhere
and
sort
of
that
also
post
drives
the
op
status
of
the
specific
interface
of
a
specific
node
based
on
things
that
are
outside
of
this
yeah.
That
I'm
good
with.
B
So
I
have
the
two
the
questions
for
comments.
The
one
one
comment
is
about.
The
this
network
status
seems
to
be
the
I,
have
data
kind
of
the
not
not
about
the
objects?
They
are
ready
or
some
stuff
I
mean
that
they
buy
it
invalid
stuff,
so
maybe
the
they.
We
should
use
this,
not
the
radio
from
the
not
ready
I
mean
the
network
is
ready,
but
also
they
are.
B
So
so
the
yeah-
maybe
they
are
I'm,
I'm
thinking
that
the
ready
is
not
the
is
not
good
to
fit
in
this
context,
because
the
user
may
misunderstanding,
the
red
network
is
ready,
I
mean
that,
so
that
this
network
is
ready
so
that
we
can
send
it
apart.
We
can
launching
the
port
understanding
in
the
pocket,
but
also
this
ready
is
not
the
ready
for
Network.
This
is
the
human
as
I
mentioned.
This
is
the
these
parameters
of
the
API
is
the
valid.
A
So
I
think
the
other
and
and
I
Rely
here
on
more
experienced
kubernetes
API
in
this
in
this
room
in
this
meeting,
I
think
the
other,
because
I'm
trying
to
reuse
one
of
the
kind
of
types
that
are
existing
so
the
other
one
that
I'm
aware
of
is
accepted.
We
can
just
use
that
instead
of
ready
right,
so
this
one
would
be
accepted.
A
Maybe
I
will
I
will
get
to
so.
Okay.
Thank
you,
I
I,
my
explanation
at
least
my
explanation
of
what
I
mean
by
the
by
behind
this
one
is
fine,
but
what
I'm
wants
to
add
is
we
want
to
go
and
maybe
then
we
can
revisit
what
should
be
naming
be.
I
want
to
go
to
the
next
one.
The
next,
a
condition
type
would
be
params
ready,
and
this
one,
as
you
can
think
of
this,
is
how
to
identify
that
my
parameters
are
ready.
A
A
My
object
expects
from
the
implementation
to
provide
the
params
ready
condition
on
this
object.
Until
then,
I
will
say
my
object
is
accepted
false.
So
that's
why
where's,
the
difference
between
ready
and
accepted
he
comes
into
Tomo
were
accepted.
I
would
imagine
accepted,
is
just
for
that
specific
object,
right,
Readiness
or
or
acceptance
just
for
like
I.
Do
a
web
hook,
kind
of
thing
right
so
I
say
accept
it.
I
accept
this
object
because
it
looks
good
I'm,
fine
with
it,
but
then
I
I.
A
That's
why
I
mean
I
want
to
use
ready
because
then
ready
identifies
additionally
that
okay,
you
additionally
specified
parameters
reference
and
your
implementation
now
has
to
say
are
those
correct
and
then,
if
they
are
correct,
send
it
back
and
tell
me
here
in
this
object
so
that
I
can
say.
Yes,
this
whole
network
is
now
ready
to
use.
A
That's
why
I
I?
That's
why
I
named
it
ready,
not
accepted,
because
at
that
point,
if
we
say
this
ready
is
dependent
on
that
params
ready,
then
I
would
prefer
name
it
ready
because
it
has
additional
kind
of
meaning
here
and
maybe
before
we
go
there.
Let's,
maybe
first
on
the
aspect
of
what
I
said
on
on
this
additional
condition.
Any
comments
on
that
is
that
something
that
is
valid
or
you
see
that
that's
not
necessary
having
the
problems
ready.
A
All
right
so
ready
is
a
response,
is,
is
the
responsibility
of
the
core
so
as
a
KCM,
probably
or
wherever
we're
gonna
have
a
controller
for
the
network?
Object
is
going
to
be
responsible
for
just
checking
basic
parameters
here
and
then
do
additional
logic.
Around
parameters.
Ref,
if
you
haven't
specified
I,
expect
a
params
ready
condition
type
to
be
in
place,
and
now
so
that's
how
the
ready
is
going
to
be
handled
now.
The
param
is
ready
is
the
responsibility
of
the
implementation
all
right.
A
A
So
what
this
will
mean
is
I
will
configure
the
whole
thing
and
your
in
your
your
implementation,
the
SRV
operator
can
say
or
or
it's
gonna
have
to
be,
probably
go
through
Motors,
but
you
will
have
everything
configured
and
as
soon
as
only
at
the
point
where
SRV
operator
is
fully
completed
when
it
applied
the
vfio
staff.
For
me,
I
configure
order
VFS
when
that's
only
completed,
then
only
then
I
come
in
here
and
say
everything's
ready.
You
can
do
that.
A
B
A
E
B
B
Thomas,
because,
okay,
let's
summarize
so
the
first,
the
please
describe
you-
you
mentioned
you
you
what
you're
saying
so.
Currently
the
the
condition
is
the
ready
and
the
Palms
ready.
So
the
you
need
to
describe
the
what
does
it
mean
and
then
also
the
you
need
to
describe
the
which
component,
the
this
capex
space,
which
component
updating
the
IDS
conditions
and
in
additions,
the
with
the,
in
addition
to
yeah
yeah,
at
least
to
me,
the
the
first
time
that
you
mentioned,
that
the
lady
is
just
the
API
Readiness.
B
So
in
this
case
the
it
it
once
this
is
explained,
then
it
seems
to
be
the
the
make
sense,
but
the
it's
not
so.
Therefore,
it's
not
intuitive
I
mean
that
the
the
user
is
just
a
user.
I
mean
that
they
they
just
have
the
several
experiment
in
the
the
current
kubernetes
at
that
time,
once
they
once
this
user
is
seeing
network
is
ready
and
then,
let's
imagine
what
they
think
about
that.
B
So
this
is
a
little
bit
the
in
just
the
comments.
So
there
sometimes
there
are
some.
Some
may
misunderstanding
network
network
status
is
ready,
means
not
the
network.
Ready
is
not
intuitive,
maybe
that
someone
is
naturally
thinking
put
the
network
status
is
ready,
means
network
is
ready,
because
this
is
ready
right.
So
maybe
we
need
to
I'm.
A
So
today,
in
in
I
think
in
phase
two,
where
we
have
that
selective
selective
availability
of
a
network
to
look
at
right
now,
we
are
not
saying
that
so
right
now,
what
we
are
saying
every
network
has
to
be
available
on
all
the
inter
all
the
nodes
of
your
cluster.
That's
the
phase,
one
only
doing
that.
A
So,
if
that
is
the
case,
what
it
means
is,
it
can
mean
exactly
what
you're
saying,
at
least
on
the
first
phase,
because
in
the
phase
two
we're
gonna
look
at
this
as
General
generic
Readiness
of
the
of
the
network
object
versus
availability
of
that
Network
object
per
node.
Okay.
Today
we
are
not
looking
at
the
second
part.
There
is
no
pair
node
availability
of
this
network.
That's
something
that
we're
going
to
look
at
next
year.
A
So
right
now
we
just
Global
Readiness
of
this
object,
and
basically
it
should
mean
that
yes,
I
can
I
could
I
could
start
and
use
that
and
reference
that
in
my
pod,
because
I
there
is
no
pair
pair
node
availability
defined
today
and
we
are
not
looking
at
it
at
this
point.
We're
gonna
look
it
later
on.
In
the
future,
so.
B
B
The
whole
whole
network
at
that
time,
yeah
I'm,
just
I'm,
just
so
yeah,
so
I
think
currency
is
not
intuitive,
that
that's
it.
So
if,
if
we
come
up
with
some
good
words
and
it's
good
and
then
sometimes
it's
not
it's
okay,
but
so
that
maybe
there's
some
user
for
user
will
finding
their
issue
once
the
this
is
a
G8.
This
is
just
concern.
A
This
is
the
the
description
of
the
Tomo
read
through
those
I
think
that
was
just
quickly
what
I
wrote
down.
Please
read
through
those
and
hope
to
describe
of
the
conditions,
but
to
my
example
previous
with
the
srov.
This
is,
of
course,
implementation
site.
This
is
where
I'm
abstracting
I
I
for
this
network
object.
It
is
still
API
based
only
and
then
how
you
define.
A
What's
behind
params
ready
it's
up
to
the
implementation
and
that's
what
I'm
saying
they
can
go
crazy
and
and
assume
Readiness
of
a
network
will
be
gated
on
whether
I
configure
something
in
a
cluster
or
I.
Don't
care
about
anything
of
this
I
will
just
validate
that.
My
params
already
specified
here
is
correct,
syntax,
wise
only
and
then
report
back.
Okay,
my
network
is
ready
and
my
parameter
and
params
already
is
true
and
then
just
to
unlock
the
other
things.
So
it's
then
it's
up
to
the
implementation
on
how
they
want
to
gate
things.
A
That's
what
I'm
getting
at
I'm
giving
I
want
to
give
that
power
for
the
implementation
to
gate.
This
object,
that's
I,
think
those
word
boils
down
to,
and
this
will
be
a
nice
API
and
I
standardized
interface.
To
do
this.
I
know
it
might
be
a
bit
complicated
for
what
you're
saying
if
I
see
where
the
network
is
ready
or
not,
but
I
think
it
will
be
exactly
that
right.
I,
look
at
the
network,
I
see
ready
to
then
I
say:
okay,
I
can
reference
this
network
in
my
in
my
workloads.
A
That's
what
it
means
and
the
other
thing
will
be
availability
of
that
Network
across
the
nodes
which
then
will
be
handled
later
on,
but
for
now
I
see
network
ready,
okay,
I
should
be
able
to
reference
it
and
it
should
work
or
the
network
ready
is
false
and
I
see
on
the
comments
parameters
ready,
not
present.
So
then
I,
look
I!
Don't
have
this
condition.
A
What's
what's
happening,
I
see
the
the
params
reference
is:
is
there
or
the
provider
who's
who's
implementing
this
so
I
go
to
that
provider
and
and
look
at
what's
the
deal?
Why
do
why
did
that
guy?
Why
did
why
didn't
that
guy?
Provided
me
perhaps
ready
condition?
That's,
how
would
I
kind
of
see
all
this
behaving
all
right,
I
will
I
think
we
are
at
the
time.
Any
other
comments
to
this.