►
From YouTube: Kubernetes SIG-Windows 20230214
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
All
right,
hello,
everybody
and
welcome
to
the
February
14th
2023
iteration
of
the
kubernetes
windows
community
meeting.
As
always,
these
meetings
are
recorded
and
uploaded
to
YouTube
so
be
sure
to
adhere
to
the
cncf
code
of
conduct
for
anybody
who's
new
here
that
pretty
much
is
boils
down
to
be
nice.
A
Let's
start
with
some
announcements,
the
first
announcements,
or
is
that
so
we're
past
enhancements,
freeze
and
the
next
kind
of
major
milestone
for
127
is
code
freeze,
and
that
is
in
one
month
from
today.
I
think
the
only
enhancement
that
Sig
Windows
is
tracking
is
part
of
the
work
that
I'm
going
forth,
that
our
Urban's
been
doing
and
I
think
we
can
maybe
discuss
a
little
bit
about
the
State
architecture,
discussion
and
the
future
of
that,
but
yeah
code
freezes
and
or
goes
into
effect
for
more
than
just
enhancements.
A
So
just
there's
changes
we
want
to
get
in
what's
working
getting
those
in.
Does
anybody
else.
A
All
right
next
is:
we
usually
give
some
space
if
there's
new
contributors
who
want
to
introduce
themselves
or
ask
any
questions
to
people
just
say,
has
no
obligation
to
do
so,
but
you're
interested
to
feel
free
to
do
so.
So,
if
there's
anybody
who's
new
on
the
call
and
wants
to
introduce
themselves
say
what
they're
working
on
they
hope
to
get
out
of
the
community
meetings.
Please
feel
free
to
just
unmute
and
say
hi.
A
B
B
A
The
agenda
I
did
want
to
discuss
something
before
we
start
with
these
agenda
items,
and
that
is
some
I
think
Sig
leadership
changes
so
for
people
who
know
our
event
has
been
contributing
in
the
in
Sig
windows
for
a
long
time,
he's
done
a
big
help
with
like
a
lot
of
the
docs
updates,
a
lot
of
driving
the
pr
and
issue
triage
and
also
working
on
the
Note
log,
Bureau
enhancement
for
for
many
years
and
I.
A
Think
because
of
his
consistent
and
like
ongoing
contributions,
we
wanted
to
offer
our
event
the
chance
to
become
a
Sig
chair.
A
So
let's
say
leads
have
already
discussed
this
and
I
think
we
all
agree,
but
part
of
the
process
is
announcing
it
publicly
and
then
there's
a
short
lazy
consensus
period,
and
then
we
can
make
that
happen.
So
I
guess.
B
A
Congratulate
our
event
on
this,
and
then,
if
anybody
has
any,
you
know,
promise
concerns,
or
you
know,
wants
to
say,
congratulations
to.
Please
either
do
so
here
or
reach
out
to
some
of
the
sick
leads
individually,
one
other
small
change
as
part
of
this,
the
kubernetes
community
is
trying
to
be
more
explicit
about
the
roles
for
chairs
and
TLS,
so
I'm
just
going
to
switch
my.
A
C
No
I'm
very
honored
and
it's
a
it'll,
be
a
pleasure
for
me
to
take
on
this
role
and
help
you
out
Mark,
so
yeah
happy
to
help
out
as
usual,
and
you
know,
keep
improving
things
in
Windows
land.
A
A
First
thing
that
I
want
to
bring
up
here,
there's
a
couple
of
issues
all
right,
so
for
anybody
who's
been
building
with
host
process
containers.
Hopefully
you
know
you
need
to
use
build
kit
in
order
to
build
host
process
containers
images
for
Windows,
because
it
there's
some
reasons
that
are
described
to
some
of
the
issues.
Well
recently,
there
were,
there
was
an
update
to
build
kit
that
updated
the
version
of
the
buildback
well
yeah
updated
the
version
of
the
buildx
plugin.
A
That's
used
to
build
things
which
caused
which
started
causing
issues
where
images
built
with
the
default
values
in
like
the
default
command
on
arguments
after
upgrading
for
build
the
build,
X,
plugin,
V
0.9.1
to
be
0.10.0
would
cause
the
images
to
no
longer
pull
if
you
were
running
container
D
on
all
versions
of
container
D,
which
is
very
unfortunate,
so
there's
some,
oh
hey,
Jay,
hey
I
will
so
we're
just
discussing
now.
A
So
right
now
the
guidance
is
to
keep
your
buildx
or
your
buildx
plugin
version
at
0.9.1,
and
there
is
a
mitigation
if
you've
already
updated.
Where
is
that
called
out
here?
I,
don't
know
if
that's
called
out
here
if
you've
already
updated
your
build
kit
version-
and
you
can't
do
this-
there's
an
extra
flag
which
I
will
put
in
here
and
in
the
slot
or
in
the
in
the
issue
and
in
the
chat
that
you
can
make
it
so
it
will
revert
to
the
old
Behavior,
but
unfortunately
it's
a
new
command
line
flag.
A
So
you
can't
add
that
to
your
build
scripts
and
have
it
work
across
versions.
Basically,
there's
been
a
whole
bunch
of
work
done
to
change
how
to
put
like
s-bomb
and
kind
of
attestation
artifacts
in
a
manifest
list
as
a
different
oci,
artifact
type
and
publish
that,
and
the
way
that
those
manifests
were
created
were
causing
some
issues
where
containerd
couldn't
actually
pull
the
image.
Because
of
the
platform
matching
logic
we
have.
A
B
So,
can
you
explain
the
the
the
how
that
what
it
was
a
bunch
of
optimizations
that
went
in
is
is
the
reason
why.
A
A
A
A
So
just
a
single
image,
not
a
manifest
list
or-
and
this
had
no
platform
matching
Logic
on
the
like
anyway,
they
said
they
said
no
like
Os
or
version
architecture,
image
or
data
on
the
container
image
so
part
of
container
d
in
order
to
help
people
make
sure
that
they're
using
the
right
version
of
the
Windows
Server
containers
on
the
right
OS
host
as
this
plan
for
matching
logic,
where,
if
your
image
has
an
OS
of
Windows
and
an
architecture
of
amd64,
it
should
have
an
OS
version
and
that
OS
version
and
containerd
checks
that
the
OS
version
will
match
match
what
the
host
is
running
on
and
if
it's
a
a
manifest
list.
A
A
Well,
they
with
Docker
within
with
an
updates
to
build
kit,
they're,
always
pushing
a
manifest
list
now
of
Type
oci
image
index
and
the
oci
image
index
is
putting
in
OS
the
OS
equals
windows
and
the
architecture
equals
AMD
64,
but
no
OS
version
on
there
and
that's
causing
the
existing
platform
matching
logic
in
container
D
to
never
be
able
to
pull
the
image
because
it
says
I
found
something
that
matches
the
OS
and
the
host,
or
that
was
an
architecture,
but
we
don't
have
a
matching
version.
So,
okay,
I.
A
Doing
with
the
host
process
containers
where
we
were
always
publishing
as
a
because
we
didn't,
because
we
could
publish
one
image
as
a
single
and
have
it
run
on
all
the
different
versions
of
Windows.
We
were
okay,
not
doing
a
manifest
list,
so
that
was
kind
of
on,
like
maybe
like
optional,
undocumented
Behavior,
but
things
broke
so
I
think
there's
some
changes
that
are
going
to
make
it
go
back
to
the
old
Behavior,
but.
A
A
Yeah
yeah,
okay,
yeah
well
Christ
CTL,
just
talks
to
like
container
D
over
the
CRI
plug-in
right
rather
than
the
whatever
like
ctrn,
because
it
does
and
yeah
you'll
see
this
in
the
cubelet.
It
I
think
it
manifests
the
slightly
different
if
you
use
the.
If
you
use
like
ctr.xz
it'll,
say
you'll
see,
there's
no
matching
platform
for
manifest
and
if
you're,
using
the
the
keyboard
or
cry
CTL
you'll
see
the
same
thing.
A
A
All
right
and
Jay,
you
came
late,
but
I
already
announced
the
our
events
being
coming
on
as
a
chair.
Oh.
B
C
C
Sure
so
we
had
a
fairly
lengthy
back
and
forth
discussion
during
the
cigarch
call,
the
main
participants,
I
guess
was
Tim
Hawkins,
Jordan,
Leggett,
Derek,
Carr
markipos
and
me,
where
I
think
most
of
the
primary
people
that
chimed
in
it
basically
boiled
down
to.
C
We
are
back
to
going
back
to
doing
what
we
do
in
open.
So
when
I
first
started
on
the
camp,
the
first
thing
they
said
is
you
need
an
API
for
them.
So
there
was
a
lot
of
back
and
forth
about
whether
this
needs
to
be
a
core
API
or
a.
We
run,
Alpha
API
and
then
Jordan
said
no.
No,
you
need
to
make
this
a
core
API
and
then
he
said
you
need
to
go
through
API
server.
C
You
can't
just
use
the
proxy
endpoint,
but
because
of
I
think
future
facing
security
concerns
and
the
amount
of
surface
area
that
you'll
be
exposing
by
doing
warlog
through
the
through
cubelet
and
Cube
cuddle.
Jordans
mine
now
is
saying
like
what
he
basically
said
is
go
back.
Let's
not
have
an
API
This
is
Not
supportable
By
Us
in
in
a
from
an
API
standpoint,
but
rather,
given
that
we
already
have
this,
you
know,
warlog
is
exposed
to
the
proxy
you're
saying
adding
more
stuff.
There
is
not
that
bad.
C
It's
it's
not
going
to
make
it
worse
from
what
it's
already
there.
The
addition
is,
we
would
disable
this
by
default
and
not
use
a
cube.
Api
server
at
all
just
use
the
proxy
endpoint.
Everything
will
live
inside
cubelet
and
Cube
cuddle
and
in
some
ways
it's
simpler,
but
it
means
that
I
have
to
rework
this
fully
to
make
it
look
like.
So
it's
basically
a
full
rewrite
at
this
point.
C
As
far
as
the
plumbing
goes
and
I
have
to
move
a
lot
of
the
validations
like
it
was
all
inside,
like
where
the
API
validation
was
happening.
I
now
need
to
move
that
into
I,
think
where
either
inside
the
cubelet
or
inside
Cube,
cuddle
or
split
it
in
two
places.
C
I
also
need
to
not
use
an
API,
so
it
means
that
I
have
to
sort
of
construct.
What
that
URL
will
look
like
inside
Cube,
CDL
and
then
it'll
be
picked
up
inside
cubelet
over
the
proxy
endpoint,
so
that
was
the
outcome
of
this
and
I've
I've
started
working
on
it.
So
a
little
bit
tricky
because
there
are
some
basic
differences
between
how
are
the
openshift
command
line.
Oc
works
from
cubecto,
but
it
should
be
doable
is
what
I'm,
hoping
and.
C
A
C
C
This
is
the
last
hurdle,
but
who
knows
but
I
I.
My
aim
is
this:
is
my
number
one
priority,
so
I
might
not
help
out
in
other
areas.
It's
my
number
one
priority
I'm
trying
to
get
this
in
before
code.
Freeze,
so
I
have
a
month.
I
should
be
able
to
do
it.
Hopefully,
unless
I
run
into
some
strange
issues,
but
I
have
help
from
Machi
also
on
the
CLI
side.
C
Okay,
so
I
should
be
help
as
in
like
he's,
he
he'll
like
he
said
he'll
review
and
you
know
help
us
get
this
through.
So
we
should
be
good.
It's
my
hope.
B
B
C
B
C
Yeah
I
I
think
what
might
even
help
is
to
have
like
clearly
defined
like
sub-architectures
inside
the
cubelet.
Maybe
that
could
be
an
approach.
That's
clearly
owned
by
you
know
like
Windows
where's,
this
rest
of
the
cubelet
that
might
even
help
but
I,
don't
know
whether
we
want
to
go.
You
know
all
the
way
to
doing
it
on
our
own,
of
course,
Michael
is
saying
yeah.
No
one
is
saying
you
can't
write
your
own.
Yes,
of
course
you
can
write
our
own.
B
A
A
A
B
Yeah
I,
don't
really
have
anything
else.
I
me
and
Damon
kind
of
started
got
up
and
running
with
some
Azure
stuff
and
we
hit
some
account
issues.
I'm
fixing
those
up
and
then
we'll,
hopefully
have
a
deployment
of
your
of
your
of
your
hack,
MD
I
added
a
bunch
of
D
like
sort
of
details
for
people
that
are
newer
to
the
cluster
API
stuff
to
it.
Okay,
do
you
Do
You,
Want,
To,
Where,
Are,
You
Gonna?
A
To
the
repository
once
for
folks
that
aren't
aware
of
what
we're
talking
about
I
think
there's
some
stuff
in
slack
but
Jay's
helping
to
set
up
some
CI
testing
or
like
a
persistent
environment,
to
run
some
CI
stuff
that
you
run
is
running
kpng
and
we're
using
cluster
API
or
kab
Z.
To
set
that
up.
Because
that's
what
the
test.
B
There's
only
there's
one
other
Deep
thing
that
I
wanted
to
update
folks,
which
is
that
so
I
was
hanging
out
with
promeita
yesterday
Monday
and
we
were
sitting
there
and
we're
like
okay,
so
the
vagrant
AWS
provider
is
deprecated,
the
vagrant
Azure
provider
is
deprecated,
there's
no
more
vagrant,
it
looks
like
and
so
I'm
trying
to
figure
out
like
what
happened
like
I'm,
Waking
I
feel
like
I'm
waking
up,
I
feel
like
Captain
America,
or
something
waking
up
in
the
iceberg
and
I'm
like
what.
How
do
people
do?
B
Virtual
machines
now
like
what
what's
the
deal
so
we're
like?
Is
everybody
just
starting
to
just
use
terraform
for
everything
now
is
that
is
that
the
way
people
spin
up
developer
environments
with
virtual
machines?
And
if
so,
we
it's
like
if
I'm
looking
at
where
we
want
to
be
for
devtools
in
you
know,
six
months
or
three
months,
it's
something
like!
B
B
We
can
do
either
one
of
those
and
either
one
of
those
is
going
to
be
a
really
exciting
and
fun
project,
and
it's
going
to
be
something
that
we
can
get
new
people
to
own
and
so
one
I'm
looking
for
owners.
If
somebody
wants
to
own
and
as
usual,
we
got
plenty
of
ideas.
B
The
question
is,
after
that,
what
do
we
want
to
do
do?
Do
we
think
we
can
do
all
of
this
with
cluster
API?
Do
we
go
off
and
build
a
Cappy
Cappy
virtualbox
provider
like
like,
and
and
do
something
where
we
glue
things
together
through
a
kind
thing,
because
I
love
this
stuff
that
Mark's
got
that
builds
Dev
clusters
with
the
custom
Coop
proxy
in
cap
C,
it's
super
useful
and
so
I'm.
Just
I
need
I.
B
C
B
B
B
C
C
Yeah
so
I
I
think
no
I
lost
my
train
of
thought,
but
what
I
was
trying
to
say
is.
B
C
C
C
Yeah,
which
is
the
tricky
part
yeah?
So
that's
where
that
that
that
issue
comes
in
right,
like
I,
remember
like
we
were
talking
about
like
using
event
kind
and
having
Windows
running
inside
a
container
that
even
might
be
possible
on
a
Windows
box,
but
like
good
luck,
trying
to
run
a
Windows
container
on
a
Linux
box,
it's
going
to
take
a
lot
of
work.
B
So
I
mean
terraform
is
the
thing
that's
the
most
flexible?
Does
it
have
any
local
story,
though,.
A
C
B
Keys
I
think
yeah
I,
don't
think
I
want
to
contribute
to
Sig
Windows.
If
that's
the
developer
story,
you
give
me
I.
C
No
I'm
just
saying
I'm
just
throwing
out
options
right
like
that
is
the
I
think
the
in.
C
D
C
C
B
B
B
C
Does
work
on
Linux,
I
I,
don't
know
about
the
Die
Hard
folks
on
like
they
would
just
want
to
use.
Kvm
is
what
I'm
thinking
yeah.
C
C
C
Yeah
so
I
see
some
questions
on
on
on
on
the
chat
from
Andy.
What
type
of
things
are
you
trying
to
run
with
terraform
I
think
what
we're
trying
to
do
is
just
bring
up
a
Windows
instance
set
up
some
kind
of
Transport
SSH
or
win
RM,
and
then
try
and
configure
the
windows
box
to
become
a
Windows
node
and
do
something
similar
for
Linux.
B
You,
the
Sig
Windows
Dev
tools
and
the
the
repo.
What
it
does
is
you
go
in
you
clone
it
down,
you
run
make
all
and
it
gives
you
a
kubernetes
cluster
potentially
built
from
source
on
your
laptop,
and
you
don't
have
to
pay
any
money
for
it.
So
it
means
that
people
can
contribute
to
you
know.
People
that
don't
work
at
Fang
can
all
over
the
world
can
contribute
without
an
internet
connection
without
a
cloud
account
we
want.
We
want.
B
We
want
Sig,
Windows,
Dev
tools
to
work
for
people
and
who,
who
don't
have
Mac
Mac
computers
and
don't
have
Linux
computers
and
right
now
we
just
don't.
We
just
don't,
have
a
solution,
a
way
for
people
to
spin
up
Windows
kubernetes
clusters
from
source
unless
they
have
a
cloud
account
or
unless
they
have
virtualbox.
C
So
yeah
and-
and
your
second
question
Andy
is-
are
the
images
hosted
anywhere
I
think
at
the
moment
there's
a
vagrant
repository
that
we
were
using
I.
Think
yeah
was
posting,
pushing
images
over
there.
B
C
A
Folks,
I've
got
a
drop
for
a
sick
note
now,
but
okay
I'm
Gonna
Make
You.
You
can
stop.
C
B
Here's
a
question
for
you:
arvind
I
mean
you
kind
of
hinted
at
this
like
what,
if
we
just
did
something
where
we
curated
a
hyper-v
recipe
and
we
curated
a
qamu
recipe
and
maybe
their
Scripts,
and
maybe
that's
fine-
you
mean
you
think
that's
doable
like.
Can
you
is
hyper-v
scriptable
in
a
in
a
reasonable
way?
I
have
not
used
hyper-v
at
all
to
be
Andy.
Do.
D
You
know
I've
only
done
a
little
bit
with
hyper-v
recently
but
effectively.
B
B
D
I've,
only
I've
only
been
recently
playing
with
hyper-v,
but
what
I
had
been
doing
was
I
used
a
Powershell
to
manipulate
hyper-v
on
a
host
and
then
the
reason
I
was
asking
about
the
images
themselves.
Is
that
effectively
you
can
build
the
image
that
you
want
to
run
and
then
you
can
run
that
within
hyper-v
I
haven't
tried
it
with
a
Linux
image,
though
I've
only
done
it
with
the
windows.
One.
B
Okay,
so
so
I
don't
know
how
often
Mandy
I
never
nice
to
meet
you
by
the
way.
I,
don't
know
how
often
you
you
come
to
the
Sig
meetings,
but
if
you're
going
to
be
here
next
week
and
you've
got
like
an
hour
or
so
do
you
do
you
feel
like
fleshing?
This
idea
out
is
something
you
might
be
able
to
do
for
us
just
to
give
us
some
seed
of
a
of
a
possible
idea
to
follow
around
this
theme
that
you've
come
up
with.