►
From YouTube: Kubernetes SIG Network Meeting for 20230330
Description
Kubernetes SIG Network Meeting for 20230330
A
A
A
C
A
A
A
Let
us
know
like
there's
some
open
questions
which
have
been
around
long
enough.
That
I'll
probably
start
closing
these
out
soon
and
be
like
all
right.
Well,
we
haven't
really
heard
back
from
you
reopen
if
you
want
to,
but
we
kind
of
have
a
way
to
go
forward
with
this
with
Gateway
API.
If
that
works
for
you,
let
us
know
what.
C
B
B
A
D
A
The
insinuation
is
that
it
might
not
actually
solve
the
problem
that
they're
originally
trying
to
solve,
but
in
lieu
of
getting
feedback
from
them
and
because
we
have
another
reason
to
do
something
similar
like
this
was
a
completely
separate
person
doing
this
for
a
different
reason.
This
might
be
a
reasonable
cause
to
just
close
the
issue
and
say
follow
up
there
or
let
us
know
if
you
really
still
need
this
in
the
as.
D
A
D
A
D
E
A
A
A
A
Yeah
they
wanted
something
that
we
talked
about
at
a
previous
meeting
as
being
vaguely
similar,
and
so
we've
left
the
open
question
of
is
this
similar
enough?
Does
the
service
backup
selector
slash
the
Gateway
API
interpretation
of
that
work
for
you
and
they
have
not
responded
yet,
but
there
was
kind
of
it
wasn't
exactly
what
they
said
they
wanted,
but
this
is
what
we
were
asking
them.
C
A
C
A
A
D
A
A
A
A
A
A
25
minutes
yeah,
so
we
could
just
take
a
look
real,
quick
because,
actually
it
probably
since
we
did
a
kept
review
last
time
and
I,
don't
think
much
has
changed.
Is
there
anything
anybody
wants
to
highlight
in
terms
of
Sig
Network
caps
today
any
attention
you
want
to
draw
to
something
or
shout
outs
that
you
want.
F
To
do
I
I
would
like
to
point
to
mind,
but
it's
not
part
of
like
release,
so
I'm,
not
sure
that
applies
okay,
the
one
that
I
pointed
out.
In
my
my
mind,
I
just
once
again,
I
make
a
shout
out
if
someone
could
take
a
look
review
and
it's
it's.
The
cap
is
a
bit
unusual
because
it's
not
a
design
per
se.
F
D
C
F
Yeah
I
had
I
had
some
sets
of
reviews
from
Dan
already
I
tried
to
address
those
and
some
other
folks,
so
I
think
most
of
them
kind
of
addressed,
but
yeah
I
would
like
to
some
other
folks
to
kind
of
take
a
look
and
and
I'd
like
to
love
this
marriage,
if
possible-
and
this
is
a
bit
unusual-
so
I'm
not
sure
how
to
treat
this
right.
It's
as
I
mentioned.
This
is
not
a
a
design.
It's
just
a
requirements
list
and
I
know
the
Caps
usually
are
are
proper
design.
F
F
Yeah
yeah,
exactly
that's
why
I'm
I
didn't
even
push
this
one
for
for,
like
I,
didn't
bother
anyone
with
this
one
for
the
like
release
and
the
code
freeze.
Last
few,
a
few
weeks
back,
just
I'm
coming
back
now
and
I
hope.
This
is
a
bit
lighter
time
when,
when
everything
is
almost
on
the
release
and
it's
there
is
no
much
of
others
priority
stuff.
So
this
is
where
I'm
trying
to
kind
of
chip
in
and
if
someone
could
take
a
look
at
this.
E
C
Yeah,
it
looks
like
you've
got
a
lot
of
a
lot
of
people
already
reviewing
this,
which
is
great,
there's
a
subset
like
if
you,
if
you
look
at
the
somewhere
down
here,
there's
going
to
be
a
set
of
owners
that
can
actually
approve
this
I.
If
you
know
it's
helpful
when
you
finally
get
to
the
list
of
approvers
that
you
already
have
some
LG
TMS
on
this
are
people
that
are
large.
F
C
A
B
There
is
not
indicated
that
the
field
was
going
to
be
deprecated
and
and
battery
is
an
annotation
and
maybe
removed
so
the
two
weeks
ago
in
the
cloud
provided
DCP
repo
someone
want
to
add
an
annotation
to
use
instead
of
this
film
and
I'm.
Trying
to
explain
then-
and
this
is
a
company
behind
that-
depends
on
these
two-
that
the
note
says
the
application
and
maybe
removed,
but
it
can
never
be
removed,
I
feel
from
an
Dae
object,
so
that
field
is
always
going
to
be
there.
B
Yes
to
demonstrate
the
this
is
a
high
consensus
in
sign
Network
that
that
field
is
not
going
to
be
removed
from
the
service.spect
of
those
voluntary
pay.
I
just
wanted
to
post
it
here
to
to
accept
if
I'm
right
or
there
are
some
concerns
and
people
should
start
doing
annotations
and
not
using
the
film
anymore.
C
A
B
B
A
G
B
A
B
For
it
yeah,
okay,
so
do
you
know
that
that
I've
been
working
in
the
service
side,
so
you
can
create
use
an
API
to
create
more
service
items
and
modify
that
option.
That
right
now
is
a
flag
date
user.
There
is
another
cap
that
that's
the
same
for
the
Clusters
type.
So,
instead
of
using
the
flux
in
the
controller
manager,
you
create
this
cluster
side
around
And,
discussing
about
that.
B
B
So
the
conclusion
is
that
having
two
API
service,
either
Across
The
Cider
with
current
API
Machinery,
is
impossible
to
reconcile
and
unless
you
have
the
status
and
use
a
central
controller
that
verify
two
apis
and
then
set
the
status
of
one
or
the
other.
But
this
still
can
be
racist
right
because
those
are
different
objects
and
and
handle
different
things.
B
The
webcal
says
is
well,
everything
will
be
easier
if
we
everything
was
in
the
same
object.
So
his
suggesting
this
Singleton
pattern,
like
a
configuration,
object
that
you
define,
set
resizers
and
port
side,
and
this
is
when
I
think
that
is
when
we
left
the
discussion.
Car
I
don't
really
want
to
add
something
else.
G
Oh
so
I
think
the
most
important
point
is
at
some
point
in
the
near
future.
Everybody
will
get
a
doc
right
should
be
short,
Doc
that
describes
the
current
situation,
the
proposition
right
and
where
we
think
we
want
to
take
this.
When
we
had
this
discussion
and
we
everybody
was
like
okay,
we
have
these
resources
that
are
shared,
such
as
cluster
cider,
herbicider,
not
ports,
DNS,
config
and
all
of
that
stuff
that
are
used
all
over
and
we
identified
like
the
user
personas
into
cluster
providers.
G
Those
are
like
the
people
who
then
the
Clusters,
be
it
cloud
or
stuff
like
ubrq,
Cube
admin
and
what
we
want
from
you
is
once
you
get
the
document.
Please
look
at
it
comment
on
it
right
because
this
is
where
we
think
we'll
drive
this
part
in
the
future.
All
right
so
comments,
questions
poking,
and
you
know
everything
that
we
do
will
be
highly
appreciated.
B
C
E
G
There
is,
there
is
a
so
it's
on
the
peripheral
of
what
we're
trying
to
do.
It's
not
directly
tied
to
it.
It's
the
idea
of
Old
Port
service.
The
idea
is,
you
have
like.
Currently
it's
really
hard
to
run
the
NBA's
on
kubernetes
network
virtual
Appliance,
you
can't,
but
you
you
do
a
lot
of
tricks
that
you
shouldn't
be
doing.
G
So
there
is
this
idea
of
all
Port
Services
that
that
has
been
discussed
and
there's
even
a
cap
for
it
that
didn't
really
move
forward
in
the
person
that
was
pushing
it
forward
to
no
longer
pushing
it
so
part
of
trying
to
do
all
of
that
is
to
figure
out
if
we
can
make
it
make
a
stepping
stone
where
those
peripheral
components
can
sit
on
top
of
something
like
that.
So
this
is
early
thinking
right
and
it
has
implication
on
the
alpha
API
that
currently
Antonio
has
out.
G
So
again,
all
we're
asking
you
to
do
right
now
is
just
to
read.
The
book
comment
ask
questions,
and
we
want
to
make
sure
that
we
covered
all
the
ends
with
this,
because
it's
gonna
be
a
serious
investment
in
terms
of
effort.
Right,
it's
not
gonna.
Look
like
a
major
change
from
the
outside,
but
internally
it
will
be
right.
So
that's
something
also.
Everybody
needs
to
to
keep
to
keep
an
eye
on
soon.
Hopefully
the
talk
will
be
up.
G
Is
relative
Tim?
Thankfully
yeah
like
it's
like
he's
like
okay,
once
we
have
the
lock,
we
should
be
able
to
have
the
data
structure,
because
that
structure
is
going
to
be
interesting
to
come
up
with
so
he's
like
he
jumped
on
on
a
test
like
yeah
yeah.
We
should
look
at
this
that
structure
deeply
and
so
on.
Again,
nothing
happened.
It's
just
an
idea.
We
want
to
make
sure
that
it's
valid
it's
sound
and
it
fits
with
all
the
peripherals
around
it.
So
yeah
soon
soon,
all
right,
soonish,
that's
it!
Thank
you.