►
From YouTube: Kubernetes SIG Network 20170713
Description
SIG Network meeting from July 13th, 2017
A
Got
a
few
items
in
the
agenda
today:
first
two
are
from
Bowie.
Do
you
wanna
kick
us
off
with
status,
41.8
yeah?
So
do
you
hear
us
yep
sound.
B
Good
boy
awesome,
so
I
put
that
as
a
generic
item,
I
noticed
a
couple
of
PRS
coming
to
me
on
these
for
ipv6
I
just
wanted
to
know
what
the
status
is.
I
miss
one
or
they're
waiting
for
notice
for
me
specifically,
but
any
PRS
that
I
think
we
need
some
kind
of
coordination.
People
are
waiting
for
review
and
so
forth.
A
Yeah
good
point:
I've
had
a
kind
of
related
question
as
well
about
how
are
we
tracking
things
and
github
at
this
point
that
are
going
into
the
milestone.
A
So
I
know
I
know
in
the
last
release
we
had
and
approved
for
milestone
label
that
went
on
kind
of
late
in
the
release
cycle.
I
know
we
should
start
labeling
things
with
that
now
so
again,
beasts
have
a
better
feel
for
for
what
are
the
things
that
are
important
for
the
milestone
versus
what
are
just
smaller
fixes?
B
So
I
think
maybe
we
should
send
out
an
email
to
the
mailing
list,
saying
if
you
have
any
PRS
or
features
that
you
want
in
1a.
Please
get
that
up
a
Hooper
milestone
for
1/8,
and
then
we
can
sort
of
understand
that
that
is
something
that
needs
to
be
done
sooner
rather
than
later
and
not
drops.
For
example,
we
could
scrub
it
to
make
sure
that
things
are
moving
along
yeah,
okay,
so
maybe
I'll
turn
down
that
email,
yeah.
A
A
B
B
A
C
Hi,
this
is
deepak
from
Bobby,
hey,
hello,
hi.
You
know
my
colleague,
Michael
he's
responsible
for
the
IP,
vs,
PR
and
I,
don't
see
Miz
Michael
are
you
there
on?
The
call
now
looks
like
is
not
there.
So
in
any
case,
I
think
I
had
a
conversation
with
them.
Yesterday
so
looks
like
last
couple
of
weeks.
We
haven't
gotten
any
feedback,
yet
it's
a
pretty
big,
PR
and
I.
Remember
in
the
last
meeting
I
was
there
as
well
and
came
mentioned
that
he's
going
to
go
through
a
very
detailed
review
process.
A
C
C
B
C
C
E
B
B
E
I
have
a
received
a
man,
Olivia
comet,
Lander
I
have
things
allow
them
and
I
need
a
yo,
Rondo
see
the
Pegasus
or
the
proposal
as
well?
Yes,
yes,
okay,.
A
E
I
think
the
better
ways
as
the
worker
in
paralyzed
by
the
you
know,
prima,
has
the
people,
many
comments
on
the
scallop.
Eva
could
appear
everywhere,
both
and
and
he
also
people
many
comments
in
the
parties
and
proposals
and
the
Kim
has
not
a
gable.
Many
some
comments
on
the
main
logical,
main
code.
Illogical
is
a
could
be,
are
so
so
so
so
so
so
the
improvement
on
the
Columbia
has
a
pending
for
some
case.
The
ad.
B
One
suggestion
is
I
know
the
code.
Pr
probably
involves
a
minimal
knowledge
of
HGVs,
but
probably
the
proposal
should
be
consumable
more
likely.
So
if
you
want
to
sort
of
get
another
round
on
the
proposal,
it
would
be
worthwhile
paying
that
out
on
the
mailing
list
and
then
I'm
pretty
sure
lots
of
people
would
be
interested
in
general
to
take
a
look
at
that.
B
C
Yeah
that'd
be
great.
If
we
can
somehow
I,
don't
know.
If
you
can
thing
I
know
you
folks
can
think
mm,
and
maybe
we
can
have
some
kind
of
code.
The
main
logic
I
think
so
that's
what
his
due
June.
The
concern
is
that
the
main
key
logic
for
this
code
base
hasn't
been
reviewed
yet
and
that's
where
the
concern
is
and
it'll
be
really
great.
If
you
can
get
it
feedback
on
that.
B
Okay,
I
mean
the
email.
Probably
could
mail
English
should
be.
You
know
hey,
this
is
a
proposed.
The
proposal
part
is
a
general
interesting,
so
we
I
think
you
know
you
may
not
need
our
PBS
knowledge,
the
IPPS
knowledge
to
review
it
and
really
want
to
overdo
it
and
then
also
click
the
code
PR,
because
I'm
pretty
sure
on
that
meaningless
are
people
with
more
detailed
knowledge
of
I
TBS.
Overall,
what.
C
Was
let
me
just
give
you
a
background,
I
think
so.
So
what
happened
was
I.
Think
when
I
think
Tim
was
on
vacation
for
a
couple
of
weeks
and
then
we
ping
them
and
we
did
not
get
any
response
back
and
then
he
came
back
in
the
last
meeting
and
he
said
not
so
so
we
didn't
want
to
bother
him
after
that
issue.
So
we
could
have
pinged
him
again.
Why
I
thought,
because
it's
a
big
PR
so
he's
going
to
it
going
to
take
some
time.
C
D
I
think
to
two
things
from
my
experience
with
Tim
in
reviewing
PRS
Percy's
been
really
busy,
and
so
you
might
just
want
to
ping
him
a
couple
of
times.
Okay,
second
season
is
like
said:
if
the
PR
is
broken
up
into
small,
discrete
chunks,
you
know
and
I
think
a
lot
of
people
do
so
there's
any
way
that
you
can.
You
know
kind
of
break
some
pieces
of
it
up.
Well,.
B
C
C
A
F
G
F
F
Then
Dan
Williams
had
a
PR
open
that
contained
most
of
this,
so
it
seemed
like
we
had
gotten
to
a
point
where
we're
least
thought
we
had
the
majority
of
what
we
wanted.
So
what
I'm
wondering
is
that
PR
only
got
closed
because
of
an
activity.
I
know
it
needs
to
be
rearranged.
A
little
bit
to
suit
the
new
earth.
F
H
I
responded
to
your
mail
like
half
an
hour
before
the
meeting,
so
I
know,
but
yeah
I
think
other
Dan's
PR
is
probably
a
good
starting
point.
The
Google
Doc,
which
I
had
written,
which
are
also
linked
to
get
a
little
bit
broader
and
I,
think
some
of
the
use
cases
and
probably
don't
want
to
look
at
at
least
four
one,
eight,
so
so
I
think
the
PR
is
probably
where
we
want
to
start
from.
H
D
Anybody
remember
what
the
final
status
so,
what
that
was
like
eight
months
ago,
whenever
we
were
talking
but
I,
seem
to
recall.
There
was
some
reason
why
we
didn't
end
up
pursuing
it,
but
I,
don't
remember
whether
that
was
just
to
keep
our
lives
simple,
because
we're
also
doing
the
network
policy
at
that
time.
Part.
I
I
I
D
To
recall,
and
not
to
like
derail
it
I
do
agree
there,
but
I
seem
to
recall
that
there
was
an
objection
raised,
that
you
can
do
that
sort
of
thing
by
means
outside
of
communities
as
well,
and
so
why
do
we
add
that
complexity
identities,
as
you
can
set
up
your
own
egress
firewall
outside
the
cluster?
Well.
I
If
it
were
a
constant
I,
let
my
recollection
of
that
discussion,
I
suggested
that
are
pointed
out.
Some
other
people
did
too
and
if
it's
a
constant,
then
sure
the
operator
can
just
set
up
whatever
constant
controls
he
wants.
But
if
you
want
to
give
kubernetes
users,
you
know
internal
admins
control
over
it,
then
the
natural
thing
is
to
put
it
in
kubernetes
objects
and.
I
H
H
B
So
this
is
a
very
mechanical,
very
simple
thing:
we're
basically
taking
all
the
sake
related
testings,
putting
them
in
same
directory,
giving
them
same
network
approvers
and
signet
reviewers,
so
that
hopefully
they'll
make
all
of
our
lives
easier.
When
we
add
tests
and
then
there
will
be
a
test
grade
entry
just
for
saving
that
word
and
we
can
just
have
a
one-stop
shop
where
we
can
see
what
the
help
of
it
is.
A
Cool
that
sounds
really
nice.
Is
there
an
action
for
people
here?
Should?
Should
people
be
volunteering
if
they
want
to
to
be
a
test
owner
or
oh.
B
F
F
J
F
Think
the
ones
that
I
was
mainly
wandering
around
where
we
talked
about
the
load,
bouncer
claim
as
well
as
pod
local
services
or
super.
They
were
definitely
interested
in
getting
them
in,
but
I'm,
not
sure
if
there
were
clear
owners
and
thus
potentially
a
not
a
fine
future
issue,
because
there's
kind
of
like
adding
it
like
a
PR
and
then
actually
defining
it
saying.
Yes,
we
want
this
to
be
an
1/8
and
adding
a
label
to
it.
So
it's
tracked
by
kubernetes
p.m.
yes.
A
So
I'm
not
aware
of
what's
going
on
with
the
load,
balancer
claim
I
vaguely
remember
us
saying
that
if
we
hadn't
got
an
owner
for
it
by
some
timeframe,
we
would
remove
it,
but
doesn't
look
like
there's
any
information,
so
then
about
that
at
the
moment.
Does
anybody
know
like
if
somebody's
working
on
that
now
or
just
intending
to
try
and
get
that
to
some
state
in
one
today.
A
Okay,
so
maybe
we
should
just
drop
that
for
now
the
pod
locality
for
services
I
think
we've
got
appropriately
at
a
priority.
I
did
talk
to
you,
the
RC
or
third
or,
however,
that's
intended
to
be
pronounced
and
and
he's
intending
to
do
some
work
on
it.
But
you
know
if
it
doesn't
make
it
then
I,
don't
think
we
need
to
divert
resources.
A
A
A
K
So
yeah,
there's
quite
a
few
us
here
or
Cisco
they've,
been
working
on
ipv6
dannion
sort
of
talk
to
you,
guys,
I
think
of
previous
meetings
in
it,
and
we've
been
doing
quite
a
few
things.
Some
has
been
just
looking
at
low-level
code
to
find
out
things
that
are
not
ipv6
ready
and
try
to
sort
of
modify
them.
Other
things
that
we've
done
is
just
Springs
run
things
and
finding
things
breaking
and
creating
bugs
for
that.
K
K
Dannion
started
on
it
a
little
bit
I'm
trying
to
help
him
some
as
well,
so
we're
trying
to
sort
of
get
our
lab
set
up
and
we've
set
up
a
docker
and
docker
to
try
to
work
with
that
and
we're
trying
to
sort
of
set
up
like
Google
cloud
engine
accounts
to
sort
of
get
things
done,
but
it
would
be
good
if
we
could
get
a
contact
of
who
to
talk
to
about
creating
that
the
end
end
test
suite
and
how
we
want
to
do
that.
I
guess.