►
From YouTube: Kubernetes SIG Network 2018-02-08
Description
Kubernetes SIG Network Meeting, February 8th 2018
A
B
So
yeah
there's
some
did
GCE
stuff
from
two
weeks
ago.
Did
anything
happen
with
that
I
know.
B
B
A
D
A
So
I
took
some
actions
last
time,
just
because
we're
some
confusion
about
what
a
lot
of
this
stuff
meant
on
the
summary
page.
So
so
up
to
some
of
the
sig
test.
Guys
and.
A
A
A
A
A
A
A
Got
a
set
of
cube,
ATM
failures.
F
That
are
all
so
I
have
an
idea
of
what
might
be
broken
there.
I
was
looking
at
the
some
of
the
failures
and
it
looks
like
it.
It's
failing
when
it
was
trying
to
install
kubernetes
ii,
and
I
and
what's
happening
is
in
coop
c
and
I
you
know
it
needs
to
create
an
op,
C
and
ice
/bin
directory
and
that
that's
failing
and
I
think
it's
failing,
because
the
debs
packages,
the
contents
of
the
deb
package,
is
wrong.
I
think
it
I,
don't
know.
F
If
anyone
knows
Deb's
packaging
well,
but
I
think
to
create
ops.
You
know
I've
been
it
needs
to
have
an
entry
for
opt
than
an
entry
for
ops,
II
and
I,
and
then
an
entry
for
ops,
II
and
I
bin,
and
that
doesn't
seem
to
be
there
and
because
that's
failing
Google,
it
depends
on
kubernetes
CNI,
so
Google
that
gets
deleted.
A
Lead
is
anybody
familiar
with
the
the
cube
CNI.
F
F
B
F
B
B
Well,
it's
not
so
much
os's
I
mean
I,
know
in
fedora,
they'd
actually
like
to
redirect
opt,
see
and
I
been
to
user
live
exec,
because
that's
where
that
stuff's
technically
supposed
to
go
per
like
the
hierarchy
standard
or
whatever,
but
I
the
thing
that
I'm
trying
to
remember
from
while
back
was
not
fedora.
It
was
I,
think
one
of
the
cloud
providers
or
something
like
that.
B
A
B
A
D
Hello
is
from
Google,
I
was
signed
up
for
accuracy
test.
We
did
look
at
it
and
I
found
a
bug
in
there
in
that
hat
myself
and
it
fix
it,
and
it
was
actually
pretty
clean
like
last
week
and
early
this
week,
and
it
just
like,
fell
in
like
yesterday,
I
think
and
so
now.
I
think
the
upgrade
firmware
is
falling
again.
If
I
actually
took
a
look
at
the
dashboard
for
the
master
upgrade
and
all
of
them
are
failing,
so
I
don't
think
it's
nothing
specific.
So
I
don't
I'm,
not
really
there's
anything.
A
G
G
F
G
Now
from
our
tests
and
spending
there's,
you
know
a
couple
of
setting
bugs
a
couple:
refactored
PRS,
that
we
still
need
to
look
at
some
usability
issues
that
were
going
through
now
and
I.
Don't
think
it
would
be
in
anyone's
best
interest
to
rush
this.
We
don't
really
gain
much
from
just
steamrolling
mr.
ga,
because
it's
not
gonna
be
the
default
anyway,
if
it
was
GA,
at
least
for
the
time
being
so
anyone
who's
been
working
on
PBS.
G
So
there
are
a
couple
github
issues
or
we
kind
of
have
a
checklist
of
things.
I
was
hoping
today
to
kind
of
create
a
big
master
issue
with
everything's.
It's
kind
of
spread
over
different
places
right
now,
yeah
I
think
that
would
will
be
the
first
step
and
then
making
sure
that
each
checkbox
is
ticked
off,
but
there's
still
a
lot
of
items
and,
like
I,
said
it's
not
worth
rushing
it
better
to
get
it
done
right,
yeah.
F
C
A
C
C
C
A
So
I
think
that's
it,
but
we,
this
I
just
one
more
thing
and
which
I
added
a
couple
minutes
ago
about
the
1.10
kind
of
cycle,
important
dates,
so
I
wanted
to
step
back
and
look
at
what
we've
decided.
We're
gonna
do
in
1:10
and
we
think
any
of
that
still
on
track
to
come
in
code.
Freeze
is
in
a
little
over
two
weeks,
I
think
yeah
just
over
two
weeks.
So.
A
F
Yeah,
what
what
we
need
is
we
need
a
CI
testbed
running,
pretty
much
I
think
the
documentation
is
ready,
but
there's
a
chance
there's
a
chance
that
it
will
be
ready
by
for
110.
But
you
know
not
not
certain
I.
Even
you
know,
I
have
have
a
test
that
showing
signs
of
life
just
need
to
debug
it
a
bit
and
then
get
it
pushed
pushed
out.
So
it.
A
A
A
F
E
F
No
we're
we're
hoping
to
leverage
what
what
you
guys
do.
I,
don't
think
we
have
anything
any
solid
proposal
for
it.
E
A
A
E
E
A
This
is
only
in
the
latter,
so
I
think
I
missed
the
meeting
where
we
defined
this.
The
spreadsheet
okay.
But
my
interpretation
of
this
is
that
it's
the
latter
and
we
probably
need
to
scope
this.
F
A
B
A
B
B
H
G
B
G
A
A
A
A
I
Just
I
just
joined,
we
don't
have
much
to
say
on
that
topic
other
than
we
have
like
a
lot
of
parallel
discussions,
multiple
threads
one
on
like.
Do
we
just
go
to
a
beach
you
one
on.
Do
we
advance
the
features
we
have
as
an
IP
block
and
egress
as
is
and
another
one
on
HTTP
network
policies
and
for
all
of
them?
There's
no
clear,
config,
I.
A
No
I
might
I
might
not
be
anymore.
I
just
do
that.
I
Okay,
so
for
this
one,
unfortunately,
we
cannot
delete
the
code
until
we
have
like
a
reasonable
way
to
let
older
clients
upgrade
and
then
downgrade.
So
the
best
that
we
can
do
was
a
PR
those
put
in
by
Jordan
to
basically
make
that
API
right
only
or
a
read-only,
so
in
1.10
it'll
be
read-only
for
the
old
extensions
thing,
but
we
can't
remove
the
code
as
far
as
I
know,
for
it
now.
I
This
is
kind
of
like
a
thing
for,
like
all
API
groups.
As
far
as
deprecating
like
I
know,
OpenShift
has
like
enough
grades.
Pickles
are
usually
far
from
that
in
our
upgrade
scripts.
We've
added
the
stuff
to
upgrade,
but
if
you
upgraded
and
then
things
went
wrong
and
you
need
to
downgrade
and
your
studies
and
those
old
things
it
gets.
A
A
J
So
I'm
kind
of
talking
for
Co
DNS,
so
we
pushed
appear
for
the
domain
party
or
is
a
transition
of
the
cube
DNS
configuration
into
a
core
DNS
configuration.
So
we
push
the
PR,
but
there
is
no
traction
on
this
PR
I,
don't
know
how
to
do.
I
know,
Bowie
you
on
the
line,
so
you
were
asking
this
transmission.
So
I
don't
know
how
can
I
already
I
to
involve.
J
A
A
H
H
H
A
I
C
B
B
A
A
A
A
I
Task
on
the
side
of
me
to
figure
out,
what's
going
on
the
crew
baby,
em
ones
in
relation
to
coober,
ladies
anywhere
I
showed
up
to
one
of
the
cluster
API
meetings
and
kind
of
brought
our
case
forward.
Should
we
delete
these
tests
of
coot
and
he's
anywhere
is
going
to
be
deprecated
or
not
used,
and
for
one
I
gather
from
that?
It
wasn't
really
up
to
the
people
on
that
call.
Specifically,
the
cluster
API
would
really
be
embedded
with
inside
of
kubernetes
anywhere.
I
It
wouldn't
necessarily
be
replacing
it
out
of
the
box,
so
I
still
have
to
reach
out
to
someone
that
cluster
life
cycle
live
paying
some
of
the
PRS.
Quite
a
lot.
I've
talked
to
some
people
that
are
trying
to
push
things
forward,
but
the
best
I
can
say
is:
if
we're
annoyed,
a
family
can
removed
from
our
dashboard
in
favor
of
cluster
life-cycle.
Looking
at
them,
but
I
still
haven't
received
any
updates
from
anyone
about
kubernetes
anywhere,
specifically.
A
A
A
A
Right
well,
I
think,
that's
it
then
we'll
meet
again
in
two
weeks.
So
that's
the
very
22nd.