►
From YouTube: Kubernetes SIG network meeting 2019-09-19
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).
A
And
we're
recording
this
is
the
kubernetes
Signet
work
meeting
for
Thursday
September
19th
2019.
Do
we
have
someone
for
issue
this
week.
B
A
D
B
F
H
J
F
L
H
B
H
B
H
E
Have
seen
something
similar,
so
there
was
a
feature
request
basically
asking
for
the
service
controller.
Don't
act
like
when
all
the
nodes
disconnect,
because
mostly
likely,
due
to
like
the
network,
partition
between
the
master
and
nodes
and
then
the
endpoint
controller
sort
of
determine
what
M
points
should
be
behind
the
low
bouncer,
and
in
that
case
the
romances
be
kept
intact.
F
B
G
F
B
H
F
D
D
A
J
B
F
That
meeting
has
come
so
I
I
started
I
just
prepped
for
this
this
morning,
I
count
1
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
open
caps
that
are
either
committed
or
in
full
request
form.
I
thought
it
was
worthwhile
to
run
through
them.
Maybe
I
share
my
screen:
no
I'm,
not
so
everybody
I
will
I'll
run
through
them.
I'll
read
off
the
descriptions
and
we
can
sort
of
touch
base
on
them.
K
K
F
H
K
H
M
F
N
F
F
Great
yeah,
so
one
thing
that
I
mentioned
somewhere
was:
we
should
think
about
for
beta
whether
we
want
to
keep
this
as
a
built
in
or
whether
we
actually
want
to
move
it
to
a
CRT
we've
kind
of
held
the
bar
for
other
topics
and
I.
Think
I
just
dropped
the
ball
and
not
asking.
Why
are
we
not
there?
You.
F
C
N
F
D
F
Think
about
it:
okay,
something
else
that
I
threw
there
that
I
wanted
to
talk
about
when
we
get
to
the
timeline
part
was
sort
of
idly
discussed.
This
idea
of
making
117
a
primarily
stability
release,
which
would
mean
maybe
we
don't
actually
advance
any
hey,
I,
see
everybody
going.
Oh
my
poor
API,
but
you
know
like
there's
some
value
in
holding
the
line
right
so
that
when
we
get
there,
the
next
one
was
graduate
ingress
to
GA.
F
Next
was
apology,
aware
service
routing
this
just
missed
the
cut
for
116
I
know
that
there's
eight
in
my
mailbox
and
I
have,
if
we're
doing
changes
for
117.
This
will
make
anyone
fun
as
well
and
also
I,
think
actually,
with
all
the
others
you
procs
and
bits
that
landed.
You
probably
should
have
a
end
way.
Conversation
about
the
intersection
of
topology
endpoints
lies
dual
stack
and
there
was
there
were
like
four
big
things
that
landed
in
qxe
and
they
all
had
to
rebase
on
top
of
each
other,
and
this
one
just
missed
the
boat.
F
F
F
J
Phase
one
and
two
are
done:
phase
three
is
being
talked
about.
We
have
a
clear
idea
about
what
we
want
to
do,
but
it's
still
not
locked
down.
Yet
I
was
waiting
for
lucky
to
finish
up
with
the
release
and
we
can
talk
and
come
out.
Come
back.
Come
back.
Come
back
was
a
sort
of
a
proposal
for
Phase
three
and
what.
J
H
K
J
F
F
C
So
I
think
the
people
who
started
the
the
kept
sort
of
abandoned
it
once
the
initial
code
was
in
because
I
was
good
enough,
but
I
was
planning.
There's
I
said
that
email
a
while
back
so
I
think.
Basically
we
need
some
CI
and
a
little
bit
of
documentation
and
then
they
should
be
ready
to
move
ahead
so
I'm
slowly.
Working
on
that
cool.
F
F
B
F
A
F
F
F
Concern
with
this.
Of
course
it
is
that
people
will
just
ignore
us
and
do
their
own
thing
until
next
quarter
and
then
they'll
just
double
balls,
and
also
that
you
know
there's
a
lot
of
nebulous
issues
that
are
open,
that
don't
have
a
lot
of
concrete
actionable
stuff.
So
it's
hard
for
the
broader
community
to
get
engaged
and
also
there's
a
lot
of
in-flight
work
and
we'd
love
to
see
me
come
to
a
head.
F
F
F
F
A
It
looks
like
we've
got
about
80
bugs
open,
labeled,
cig
Network.
Maybe
we
could
just
set
a
explicit
goal
as
a
snake.
To
focus
on
that
is
kind
of.
One
of
the
headline
features
doesn't
necessarily
have
to
mean
stopping
other
efforts,
but.
H
M
J
J
J
F
G
D
F
D
F
A
Cool
thanks,
Bowie
thanks
Bowie
next
billion
gender
was
doesn't
have
a
name
next
to
it,
but
Debbie
and
Buster
and
NF
tables
bug
you
damage
it.
F
Well,
we
can
talk
about
it,
for
anybody
who
wasn't
following
along
Debbie
and
Buster
is
busted
on
Quran
T's
by
default,
because
the
new
iptables
1.8
point
X
uses
n
of
tables
under
the
covers,
but
the
IP
tables
built
into
the
Q
proxy
container
uses
the
old
tables
interface
and
the
kernel
does
not
watching
mixing,
wait
and
there's
no
really
good
way
to
detect
between
them.
Although
Dan
posted
a
pack
tacular
script
that
seems
to
do
the
job,
that
was
a
compliment.
C
I
think
so,
and
so
there
are
two
problems
you
know
what
one
is
that
there's
also
a
bug
in
IP
tables
182,
which
Debbie
Buster
you're
shipping.
So
but
then
it's
also
the
mix-and-match
problem
and
the
mix-and-match
problem
is
especially
a
problem
because
we
are
building
all
of
the
official
packages
on
Debian
I.
Think
it's
Sid,
whatever
the
1b
or
buster,
is,
and
that
only
has
IP
tables,
1
6.
So
there's
no
way
we
can
make
it
support
the
the
N
F
table
stuff,
and
so
that's
that
sort
of
complicates
possible
solutions.
So.
F
F
So
we
need
to
get
somebody
who
understands
that
to
help
make
sure
that
I
didn't
screw
it
out,
but
I
was
able
to
successfully
build
a
debian
base
muster
and
then
an
Ivy
tables
on
top
of
that
on
muster
when
I
didn't
do
yet
was
start
to
patch
in
the
script
that
you
wrote
like
basically
going
and
were
ex
post
facto
replace
all
the
IP
table,
binaries
with
a
script
that
runs
your
your
heuristic
and
then
decides
which
other
IP
tables
version
to
run.
I
catch
I
hand
him
down
this.
F
G
F
F
A
Okay:
next
up,
we've
got
ingress,
v1,
API
working.
O
D
Yeah
so
Chris
I
see
you're
on
the
call
as
well
so
Chris
and
I
were
talking.
We
were
basically
ingress,
v1
GA.
We
want
to
kind
of
get
a
good
cadence
going
and
we
were
thinking
that
we
would
start
up
just
the
meeting.
I,
don't
know
what
they
qualify
for
working
group.
They
say
every
week
with
input,
interested
ingress,
d1
implementers
and
try
to
sort
of
wrap
up
the
cap
and
then
get
all
the
implementations.
D
F
Chris
I
know
there's
the
topic
around
the
error
message
around
the
renamed.
You
feel
and
I.
Don't,
unfortunately,
don't
have
a
good
update
for
you,
except
that
I
spoke
with
the
API
machinery.
Folks
who
said
yeah.
Actually
the
validation
should
be
running
on
the
version
sites
and
if
somebody
wants
to
go
dig
into
the
API
machinery,
I
have
some
starting
points.
It's
probably
not
a
short-term
project,
but
it
would
be
good.
So
if
anybody
here
feels
like
cross
pollinating
with
a
guy
machine,
let
me
know,
and
happily
the
point
there
went
away.
F
I,
don't
think
it's
a
blocker
I
think
we
can
just
do
something
dirty
like
update
the
strings
when
it
says
both
names
or
something
and
leave
it
to
do
and
come
back
to
it.
I'm
trying
to
motivate
the
again
machinery
team
to
do
this
themselves,
but
I
did
start
to
peel
back
the
layers
of
the
onion
and
it's
brain
smells.
F
Q
H
I
I
have
basic
stupid
questions
here,
so
we're
talking
about
advancing
from
V
1
beta
1
to
V
1,
but
making
big
changes.
Why
is
it
not
going
to
be
1
beta,
2.
F
D
Q
Q
B
I
I
I
D
M
M
M
D
D
D
Two,
so
I
think
the
working
group
will
initially
start
out
with
v1
just
to
figure
out
what
to
do
there
in
terms
of
scope
and
then
probably
the
same.
Interested
people
will
end
up
rolling
into
the
v2,
but
initially
I
just
wanted
to
get
the
v1
just
kept
for
the
view
ones
already
out,
and
we
already
made
some
progress.
There
is
to
just
wrap
up
and
figure
out
like
what
specifically
we're
going
to
do
there,
but
the
the
kind
of
people
interested
will
be.
Basically
the
same
I
would
expect
and.
F
F
F
F
F
M
R
R
F
V
1
should
be
good
enough
so
that
we
don't
have
to
spend
any
more
time
or
attention
on
the
grass
and
we
can
focus
our
energy
revolution.
What
I
like
about
this
new
proposal
is
it's
a
much
broader
vision
for
load
balancing
so
comes
with
commensurate
complexity,
so
people
please
go
look
at
it.
Think
about
it.
Is
it's
missing
a
lot
of
stuff
like
it's
the
skeleton
of
the
proposal,
but
there's
a
lot
of
the
flesh
that
isn't
there
yet
make.
A
A
B
A
A
The
next
one,
those
for
me,
we
don't
talk
much
about
this
main
thing.
I
want
to
just
point
out
that
the
tentative
schedule
her
117
is
being
figured
out.
It's
a
slightly
shorter
cycle,
the
others
as
usual
towards
the
end
of
the
year.
She
can
check
out
the
skeptical
there
I've
already
talked
a
little
bit
about
what
we
might
want
to
do
in
that
timeframe.
A
F
F
F
We
don't
have
to
decide
here
think
about
it
if
you're
gonna
be
in
San
Diego
anyway,
it's
a
low-impact
way
of
getting
some
funds
data
tying
and
talking
to
people.
The
slides
that
produce
for
the
same
eating
are
probably
a
lot
less
velocity
than
the
slides
you
produce
for
a
normal
session,
I'm
willing
to
help
with
slides
and
presentation
and
all
you
know
whatever
people
need.
If
you
are
anxious
about
present
I'm,
also
happy
to
help.
F
S
F
F
D
Q
F
R
R
B
F
F
M
B
F
M
B
Broadly,
with
sig
usability
we're
looking
at
some
of
the
defaults
that
are
basically
unfriendly
and
trying
to
see
if
it
would
be
possible
to
nudge
them
in
like
a
more
turnkey
situation
like
default,
close
sources,
people
open
network
makes
a
lot
more
sense.
If
you
want
to
open
it
all
up,
then
you
pawn
can
be
big.
I
know
what
I'm
doing
conveyed
that
kind
of
thing.
Interesting.