►
From YouTube: Kubernetes SIG Windows 20170919
Description
Kubernetes SIG Windows 20170919
A
I
hi
everybody.
This
is
another
meeting
for
for
sick
windows.
Welcome
to
our
channel
I'm
gonna,
kick
things
off
and
give
everyone
an
update
on
a
couple
of
different
work.
Streams
are
happening
on
the
sick
windows
community
and
then
I'll
will
open
up
the
floor
for
questions
comments
as
well
as
any
other
thing
that
someone
wants
to
bring
up
so
on
the
work
streams.
There's
three
separate
work
streams
going
on.
A
At
the
same
time,
the
first
one
is
to
have
folks
from
aprender
that
have
actually
started
looking
into
cube,
ATM
and
essentially
we're
making
investments
into
a
particular
tool
to
allow
it
to
stand
up
communities
cluster
and
also
add
Windows
nodes
into
the
environment.
There's
going
to
be
a
significant
net
benefit
for
our
community
for
two
reasons:
number
one
is
going
to
significantly
simplified
the
steps
of
adding
a
Windows
note
to
your
cluster.
A
So
so
now
it's
going
to
be
just
a
couple
of
simple
commands
and
then
you
don't
have
to
run
those
huge
scripts
that
we
have
today
and
the
second
thing
is
going
to
provide
the
basis
for
us
to
create
automated
tests.
On
top
of
the
seafood
in
those
work,
so
those
are
the
tests
that
are
gonna
run
nightly
to
make
sure
that
neither
the
kubernetes
core
project,
neither
any
one
of
our
own
PRS
breaks
things.
A
So
that's
probably
the
automated
something
that's
going
to
happen
post,
but
you
know
for
now
we'll
get
the
cube
ATM
work.
The
second
work
stream
is
the
folks
from
cloud
base
are
working
and
they're,
basically
making
good
progress
on
the
CNI
plugin
for
overlay
Network,
actually
working
on
two
separate
work
streams
here,
a
number
one
is
the
I
use
the
Windows
Server
2016
rs3
build.
A
That's
gonna,
come
out
sometime
soon
to
create
a
virtual
overlay
network
using
ovn,
OVS
and
they're,
also
having
a
second
work
stream
that
will
support
an
overlay
network
without
the
requirement
for
our
s3.
Now
this
each
each
of
those
solutions
has
its
own
limitations
as
well
as
advantages.
If
our
s3
is
being
used,
you'll
be
able
to
have
multiple
containers
per
pod
and
it's
gonna
use
a
shin-hye
plugin
for
Microsoft.
A
If
our
s3
is,
you
only
have
only
one
container
per
pod
did
a
couple
of
pieces
to
see
if
there
was
a
proof
of
concept
or
a
solution.
If
we
do
require
RS
3
to
a
multiple
containers
per
pod,
that
was
a
hacky
solution,
so
we
opted
out
for
using
the
Microsoft
implementation
and
using
rst
as
the
basis
for
that
multi
container
per
pod
solution.
A
Now
Microsoft
is
also
working
on
its
own
set
of
PRS
and
Ganesh's
is
going
to
call
so
I'll.
Let
him
update
us
on
that
at
the
high
level,
Microsoft
is
working
on
basically
in
those
peers
into
1.9
of
kubernetes
and
we're
gonna
start
talking
to
the
networking
guys
from
SiC
network
to
start
pushing
the
ball
forward.
Dinesh
wanna
give
that
update
really
quickly
sure
so.
B
There
are
two
peers
we
are
working
on
one
as
so
today.
Windows
doesn't
have
the
queue
proxy
kernel
move
support,
so
a
so
far
s3.
We
basically
enable
the
load
balancing
support
in
in
the
platform.
This
is
the
same
data
pane
trimer
that
use
an
usher
and
or
Sdn
and
Raman
the
PFP.
So
we
are
bringing
in
a
cube
proxy
kernel-mode
support.
That
is
one
PR
and
the
other
PR
is.
B
We
are
basically
bringing
in
native
CNI
plugin,
with
which
you
can
do
multi
shed
multiple
containers
in
a
pod,
and
then
basically,
we
had
some
lot
of
hacks
in
the
2016,
where
you
have
to
have
multiple
network
adapters
to
the
container
in
order
to
provide
Internet
connectivity.
So
we
are
removing
that
workaround
and
basically
trying
to
map
the
exact
set
of
features
that
linux
has
today
to
Windows
platform
and
then
we're
trying
to
bring
things
negatively.
So
that
is
the
second
PR.
A
Thank
you
Dinesh,
so
we'll
all
work
together
and
figure
out.
You
know
how
all
these
significant
investments
from
Microsoft
come
make
it
into
upstream
kubernetes.
That
way.
The
combination
of
the
two
peers
at
the
national-
they
are
s
rebuilds
from
from
Windows,
and
the
work
that
we're
doing
from
cloud
based
are
gonna,
get
us
into
a
point
that
you
can
actually
have
a
really
solid
beta
solution
for
our
customers.
Now,
if
all
the
peers
from
Microsoft
land
in
one
point
we're
gonna,
have
to
figure
out,
do
we
get
our
bed
out
with
1.9?
A
Or
can
we
get
the
bed
out
earlier
right?
Now
we
can
have
that
discussion.
Our
goal
was
to
get
a
beta
before
Kubik
on,
but
I.
Don't
I
think
it's
too
early
right
now
to
tell
let's
we're
gonna
get
the
PRC
in
make
sure
everything
stabilize
make
sure
cloud
base
has
enough
time
to
bake
their
solution
on
top
of
it
and
then
we'll
make
a
good
decision
for
beta
I
know.
A
The
so
the
the
other
update
I
wanted
to
have
was
essentially
so
so
we
have
cloud
base
working
on
this
who
have
Microsoft
on
this
there's
also
a
tiara
and
calico.
They
are
looking
into
the
CNI
and
the
PRS
from
Microsoft,
and
potentially
you
could
have
another
CNI
coming
from
them.
That
could
support
the
Windows
effort
right
now.
A
We
don't
have
any
solid,
committed
plans
from
them,
so
we're
just
taking
this
on
a
on
a
on
a
day-by-day
on
the
on
the
hosting
front,
given
that
rs3
is
gonna,
be
a
major
requirement
for
hosting
multiple
containers
per
pod
Peter
as
well
as
you
know,
obviously
the
guys
from
Microsoft
Antoni.
How
is
not
here
on
the
call
today,
but
I'm,
assuming
you
guys,
are
engaged
with
Microsoft
to
make
sure
that
one
others
three
ships,
you
support
it
on
Google,
Cloud,.
A
Yeah,
you
don't
need
to
bring
any
commitments
or
say
anything
more
than
that,
so
that's
good!
Now
our
goal,
you
know,
given
that
all
the
instructions
that
we
have
and
most
people
have
been
trying
this
on
either
riser
on
Google
cloud
when
I
make
sure
that
both
platform
support
the
bill.
That
way,
you
know
we'll
be
able
to
hit
the
ground
running
once
it's
available.
Yes,.
C
A
Excellent
and
actually
on
the
windows
inside
they're
built
there's
a
couple
of
bugs
that
cloud-based
have
filed
and
the
Nash
and
his
team
are
working
on
fixing
those
they
are
not
in
the
latest.
Insiders
build
that
came
out
maybe
ten
days
ago,
but
there's
they
should
be
on
the
next
available
one.
So
so
so.
B
A
A
Excellent
yeah,
that's
awesome,
then
David
Porter,
the
internet,
the
intern
from
Microsoft.
That's
the
last
I
think
that
we
have
has
completed
the
PRS
for
for
basically,
he
has
a
combination
of
two
PRS
that
he
had
created.
The
first
one
was
on
the
node
stats
and
the
other
one
was
for
the
pod
level
stats.
The
node
studs
was
approved.
A
It's
not
completely
approved
yet,
but
it
should
be
approved
and
should
be
merged
soon,
after
so
node
stats
approved
and
it
should
be
merged
as
soon
as
the
queue
is
open
for
1.9
and
the
pod
level
starts
soon
after.
If
you
guys
are
interested
in
hearing
more
about
learning
more
about
these
PRS,
there
are
pr5
1152
and
pr5
zero.
Three
nine
six.
A
D
A
We
actually
volume
secret
and
config
maps,
we
already
tested
them
and
there's
a
very
nice
document
on
how
to
configure
those
there's
a
couple
of
small
issues
around
secrets,
just
because
there's
no
a
synchronous
swap
on
Windows
but
Bob
and
the
team
were
able
to
find
a
few
workarounds
I
believe
I'll
ask
Bob
to
post
his
document
on
the
communities
everybody
concede
and
he'll
posted
on
sick
windows.
Shortly
after
this,
he
wrote
a
nice
document
explaining
how
people
can
leverage
this.
That's.
A
D
A
You
cool
I,
don't
know
how
many
of
you
guys
are
going
to
end
up
going
to
a
Kubek
on
in
Austin
a
short
drive
for
me.
I
live
in
Houston,
so
I
should
be
about
together
fairly
easily,
but
we
did
submit
a
session
for
our
work
around.
You
know,
windows
and
networking,
and
all
these
things
they
told
us
that
they'll
will
be
notified
within
the
next
week
or
so
if
our
session
is
approved,
but
hopefully
we
do
get
the
session.
A
Alright,
that's
it
I,
guess
short,
meaning
today,
I'll
upload
it
on
YouTube
like
always
feel
free
to
engage
with
us
in
any
way
possible.
We'll
have
to
get
more
more
people
and
more
ice
on
this
work
and
I'm
super
excited
about
rs3
and
all
the
other
work.
That's
happening
on
the
network
inside
I.
Think
it's
a
tremendous
improvement.
It's
getting
us
closer
to
parity
with
Linux
and
people
will
be
able
to
put
production
workloads
on
Windows,
Server
containers
and
open.
It
is
very
simple:
cool
have
a
great
day.
Everyone.
Thank
you.