►
From YouTube: Network Policy API Bi-Weekly Meeting for 20201201
Description
Network Policy API Bi-Weekly Meeting for 20201201
A
A
A
A
A
B
Go
ahead,
yeah
was
just
hacking
on
looking
just
looking
at
it
with
abhishek.
I
think
we're
we're
just
got
to
figure
out
one
thing,
and
that
is
that
I
gotta
compile
I'm
doing
this
right
now.
I
gotta
figure
out
how
to
compile
the
code
to
figure
out
it
feels
like
the
api
machinery.
Folks
are
implying
that
there's
some
kind
of
schema
on
read
that
we
can
do.
B
Go
like
you
know,
is
defaults.go
called
on
read
or
only
on
right
is
really
my
question,
and
I
don't
know
if
anybody
here
can
answer
that,
but
that's
going
to
determine
whether
we
can
do
the
solution
we
proposed
in
the
kep,
which
is
really
simple,
which
is
just
modify
defaults.go
so
that
you
know
it
always
adds
that
label
and
if
we
do
that,
we're
fine
and
we
don't
have
to
do
anything
else,
because
we
don't
really
even
have
to
worry
about
going
and
setting
old
name
spaces,
because
anytime,
someone
uses
the
api
server
to
read
something
it'll
automatically
get
that
label
put
on.
B
That's
pretty
much
the
only
question
we
have
to
answer.
Everything
else
is
just
nets.
C
B
B
B
I
think
the
only
thing
we
need
to
worry
about
is:
can
we
get
that
schema
on
read
thing
working,
because
if
so
we're
fine
right?
Because
anytime,
you
read
the
value
you'll,
get
that
label
right
and
whether
or
not
old
records
in
etcd
have
that
information
or
not
winds
up
being
irrelevant
and
in
fact
the
idea
of
whether
even
that
label
is
written
to
etcd
seems
like
it
would
be
irrelevant
too
right.
E
Okay,
so
I
think
the
one
thing
that
I'm
a
little
bit
you
know
yet
unclear
on
is
how
to
handle
the
versions
cube.
For
example,
you
have
multiple
api
servers
offering
different.
You
know
ones
upgraded
the
other
one
system,
the
other
one
is
not
so
in
this
case,
if
a
read
is
being
performed
or
served
by
the
older
api
version,
it
doesn't
have
the
it
doesn't.
E
Have
the
you
know
the
labels
writing
capability,
so
that's
the
only
thing
that
we
need
to
ensure
that
doing
it
in
the
fall
sort
go
versus
doing
it
in
you
know,
strategy
like
the
prepare
for
or
prepare
for,
update
or
create
whether
that
will
solve
the
problem.
I
think
jordan
was
kind
of
alluding
to
the
fact
that
if
we
use
the
latter,
it
might
solve
the
case
for
both
versions
queues.
E
B
Yep,
so
I'm
gonna,
I'm
just
gonna
see
you
know
the
thing
is
ricardo.
I
just
got
to
see
yeah
like
what
happens
when
we
actually
implement
this
change
like
what
do
the
objects
look
like
right?
We
got
to
see
it
and
the
biggest
question
winds
up
being.
Is
there
some
hack
we
can
put
in
there?
That
makes
it
so
that
old
objects
that
were
there
before
we
started
have
that
label.
You
know
and
I'm
assuming
there's
something
in
api
machinery
that
allows
you
to
do
that,
a
last-minute
modification,
something
you
can
do.
A
B
F
A
E
Yeah
last
week
was
a
short
one,
so
we
canceled
the
meeting.
We
should
be
resuming
this
week
so
hopefully,
next
week,
we'll
have
a
better
update
on.
A
A
Okay,
so
about
the
part
ranges
I'm
I'm
just
waiting
for
some
sort
of
improvement.
I
think
that
andrea
andrew
is
not
here
today.
He
he
told
me
that
he
was
going
to
to
take
some
time
off
this
week,
but
I
think
that
we've
had
we
have
reached
some
some
agreement
about
the
part
range
I
just
can't
see
if,
if
we
got
the
agreement,
that
exceptions
can
be
dropped
right
now,
so
I've
I've
run
I've.
I
I
have
tried
to
search
about
my
my
security,
the
security
policies
of
my
company
and.
A
In
my
case-
and
I
have
like
I
work
in
a
pretty
restricted
environment,
so
the
default
blocking
policy
covers,
like
eight
parts,
mostly
of
them
are
regardless
file
sharing
and
and
the
remote
common
like
rsh,
that's
those
old
ones,
extern.
So
having
eight
port
ranges,
I
think
that
it
is
plausible,
even
if
we
put
them
in
the
cluster
scoped
network
policy
later
like
in
the
future.
This
is
the
network
policy.
A
If
you
you
can
allow
anything,
but
those
parts
like
having
those
as
as
a
part
ranges
as
a
support
range
feuds
already,
like
eight
arrays,
so
I'm
just
waiting
for
four
last
comments
from
team
and
andrew,
and
I
think
hopefully
this
is
going
to
get
merged
and
we
can
start
I.
I
will
probably
start
working
into
this
also,
so
I
can
like
take
a
look
into
how
he's
developing
api
machinery
stuff,
even
if
this
is
just
adding
a
new
field.
A
C
Hi
ricardo
satish,
so
yeah,
so
we
I
mean
I
don't
have
any
context
about
the
service
selectors
but
but
happy
to
pitch
in
there.
Like,
I
mean
happy
to
take
it,
I
mean
I'm
actually
pretty
new
to
this
group,
so
I
don't
have
any
context
about
the
service
selectors
and
what's
the
motivation
behind
it,
but
with
enough
information,
provided,
I
think
I'll
be
able
to
pitch
in.
B
We
definitely
need
an
owner
so,
like
we've,
we've
had
a
few
people
sign
up
to
own
this
thing,
but
nobody's
really
actually
done
the
work
of
owning
it.
So
I'll
say
what
I
always
say,
which
is
that,
like
anybody
in
this
group
will
totally
help
you,
but
you
know
it's
just
like
anything
else.
It's
like.
We
just
need
someone
to
really
own
it.
You
know
what
I
mean
and
and
and
make
it
happen.
B
You'll
have
all
the
support
you
want
like
technically
and
you
could
get
on
the
phone
with
me
or
ricardo
or
abhishek
and
we'll
help
you
with
whatever
you
need
to
do
and
give
you
context
for
it
and
everything
else.
It's
just
a
matter
of.
Do
you
have
the
time
to
dedicate
to
it?
You
know
what
I
mean
to
really
making
sure
that
it
happens.
A
A
E
I
think
there
are
a
couple
of
other
user
stories
also
that
needed
some
owners.
So
you
know
whatever
you
feel
more
comfortable
with
or
you
you
think
that
is
more
important
like
there
is
no
the
the
ability
to
select
nodes
or
apply
policies
for
nodes.
The
service
selector
is
one
and
then
I
think
ricardo
has
shared
the
document
which
has
a
few
more
user
stories.
E
C
B
G
B
B
It's
gonna
be
a
fun
fun
project
for
sure.
Are
you
familiar
with
reasonably
familiar
with
network
policies
and
stuff?
Have
you
used
them
before
or
do
you
need
help
getting
started
with
them?.
C
Yeah,
so
I've
used
them
before
I
mean
I'm
actually
a
part
of
the
gk
networking
team
here
at
google,
okay,
so
yeah.
I
do
have
like
some
context
about
the
network
policy
and
yeah.
B
A
A
B
B
B
I
did
I'm
so
the
only
thing
I
really
modified
was
I
modified
the
package
apis,
or
can
you
put
a
better
coloring.
B
B
A
F
B
B
Yeah
so
anyways
for
those
of
y'all
that
are
oh,
I
know
what
it
is.
B
I
think
it's
that
something
is
wrong
with
your
dns
input
get
grab.
Something
is
wrong
yeah.
I
remember
this
and
the
reason
I
remember
this
is
because
yeah
there's
something
when
you
run
hack
local
up
cluster,
that
you
need
to
have
the
coop
dns
stuff
set
up
correctly,
but
you
know
what
we
don't
need
that
right,
so
we
actually
don't
have
to
exit
here,
because
we're
only
testing
one
really
stupid,
simple
thing.
So,
let's
just
see
what
happens
if
we
run
this
without
that
exit?
B
B
B
B
B
A
B
B
B
B
B
I
know
I
don't
know
man,
I
don't
know
why
I'm
doing
this
I
well
so
the
reason
is
because
my
it's
not
calibrated
somehow
so
once
I
started
screen
sharing.
Have
you
ever
had
that
weird
thing
where
you
get
like?
Where,
like
you,
try
to
copy
something,
and
it's
like
three
lines
above
I'll:
try
it
anyways,
though
okay,
actually
it
is
working
now,
so
I
can
just
copy
paste
it.
Okay,
you're
right.
D
B
B
Yeah
the
old
name
spaces,
I
guess
so
yeah.
E
I
think
the
only
thing
that
yeah
the
defaults
go.
It
seems
like
it's
working
right.
I
think
the
only
concern
is
that
the
mixed
kpi
server
versions,
like
the
one
which
does
not
have
this
code,
and
if
that
is
serving
some
name
spaces,
then
how
does
it
react
to
that
and
from
jordan's
comment
it
seem
like
if
we
use
the
prepare
for
update
method,
we
could
potentially
address
that.
E
This
is
what
tim
suggested
and
jordan
kind
of
agreed.
E
Yeah
but
but
he
didn't
leave
him
comment
on
the
the
version
skew
section
which
kind
of
handles,
which
kind
of
is
satisfied
by
doing
the
changes
in
the
strategy.go.
D
A
Yeah
great
I'm
going
to
start
hacking
into
the
end
part.
So
as
soon
as
this
is
also
approved
in
the
cap,
we
are
going
to
merge
it,
probably
as
often
v121
yeah
great
job
changes.
So
I
feel
like.
Let
me
see
also
if
someone
wants
to
to
pick
this
up
like
the
end
part
when
this
is
a
prove
it.
I
I
don't
care
like
not
having
this
as
my
code,
so
please
join
me.
B
Okay,
so
I
mean,
I
guess,
that's
we
can
kind
of
close
out.
I
mean
as
far
as
I'm
concerned.
That's
that's
all.
I
got
abhishek
ricardo
anybody
else.
If
you
all
want
to
maybe
stay
on
after
this
and
talking
what
we
can,
I
I
think
abhishek
ricardo,
and
maybe
you
can
just
I
think,
abhishek
mentioned
there's
a
couple
things
he's
going
to
clean
up
and
then
I
can
clean
up
this
part
of
it
and
then
I
feel
like
we're
pretty
close,
but
I'm
happy
to
hang
out
more
if
y'all
want.
H
Hey
everyone
I
j.
I
would
just
like
to
introduce
myself
sorry
I
hopped
on
late.
I
couldn't
figure
out
the
meeting
password,
but
I'm
andrew
I'm
from
red
hat.
I'm
a
software
engineer
at
red
hat
and
I've
been
working
a
lot
on
network
policy
and
upstream
of
in
kubernetes,
so
thought
I'd
hop
in
here
and
try
to
help
out
where
I
can
so
I'll
should
be
around
from
now
on.
B
Oh
cool
andrew:
we
we've
worked
with
some
folks
over
at
red
hat
before
for
sure
we
we
were
working
on
some
network
policy
stuff
with
antonio
and
stuff.
Are
you
in
that
group
or.
H
Yeah,
I'm
pretty
adjacent
to
them.
I'm
working
with
the
openshift
networking
team
right
now,
but
I'm
part
of
the
octo
team,
so
kind
of
kind.
B
B
Yeah
we,
you
know,
we
have
this
policy
validation,
suite
that
we
were
just
running
on
ovn
kubernetes
about
three
weeks
ago
and
we
found
a
bug
and
we
helped
you
all
fix
it
and
right
so
I
need,
but
I
need
help
getting
that
finished.
The
reviews
finished
on
that
pr.
So.
A
Okay,
so
I
think
that's
it
for
now,
folks,
we
are
in
the
end
of
the
agenda
today
and
see
you
all
next
week,
next
monday.
Thank
you.
Everyone
stay
safe.
All
right,
see:
y'all
yeah
thanks.
Everyone.