►
From YouTube: Kubernetes SIG Windows 20200414
Description
Kubernetes SIG Windows 20200414
A
Hello,
everybody
and
welcome
to
another
sick
windows
Meetup.
This
is
the
14th
of
April
and,
as
always,
there's
a
recorded
meeting.
So
please
adhere
to
the
CN
CF
code
of
conduct.
I
want
to
actually
start
by
telling
you
guys
one
very
quick
thing:
YouTube
creators
added
or
has
changed
how
it
works.
So
a
lot
of
our
community
meetings
for
the
last
three
weeks,
even
though
I
have
them
recorded
and
I
have
them
upload
on
youtube.
I
cannot
link
them
to
our
playlist
that
we
have.
A
The
kubernetes
team
is
trying
to
fix
that
because
it
impacted
pretty
much
everything.
So
the
way
it
works
is
that
there's
a
the
playlist
is
owned
by
kubernetes
and
then
I
was
added
as
an
editor
I
do
the
new
with
a
new
freaking
editor
studio.
You
cannot
link
to
playlist
that
you
don't
have
ownership
of,
and
you
know
first.
B
A
The
workaround
that
you
saw
before
is
you
told
them
that
there's
a
missing
feature
and
they
give
you
a
link
to
go
back
to
the
old
creator,
studio
and
you're,
able
to
add
them,
but
that
link
no
longer
exists
actually
got
an
email
saying
you've
been
using
that
it's
no
longer
available,
so
so
I
don't
know
what's
gonna
happen,
apparently
there's
some
automation
of
the
kubernetes
contributors
experience.
People
are
working
on
and
it's
gonna
fix
this
thing.
A
A
C
Yeah
I
added
this
one,
so
it
looks
like
quad,
is
pretty
close
to
having
all
or
has
all
of
the
test
images
working
with
others,
virtual
back
onto
wait
as
nano
server
images
and
I
think
that
were
just
want
to
I.
Think
we
there'd
be
a
lot
of
benefit
to
moving
all
of
the
images
over
to
nano
server
from
speed
and
just
being
able
to
run
on
smaller
nodes
with
smaller
discs.
C
D
D
I
had
one
question
so
I
recall
and
I
was
just
trying
to
pull
up
the
doc
link.
I
couldn't
find
it
right
away,
but
it
used
to
be
or
I
used
to
be
documented
that
nano
server
containers
would
only
run
on
the
semi-annual
channel
versions
of
Windows
and
they
wouldn't
run
on,
say
Windows
Server,
2000
19.
Is
that
still
accurate?
So.
C
I
think
that
the
they
they
would
run
on
Windows,
Server,
2000
19,
but
I
believe
that
they
were
going
to
have
the
same
support.
Cayden
ours
like
Support
Lifecycle
as
the
the
sack
images
so
the
18
months
aboard
cycle.
We
are
working
with
teams
inside
of
Windows
to
make
sure
that
that
gets
extended
for
Windows
Server
2000
19.
That
is
a
concern
that
that
we
had
and
I
believe
that
right
now
we're
thinking.
C
We
wanted
to
try
this
for
a
little
bit,
while
the
images
are
still
supported
to
see
how
much
it
can
improve
everything
like
how
much
faster
the
test
passes
are
and
come
back
with
some
some
numbers
for
that.
All
of
the
work
that's
been
done
to
make
the
test
images
be
able
to
run
on
nano
server
would
accrue,
but
they'll
still
run
fine
on
Windows
the
server
core
containers.
So
if
we
need
to
switch
it
back,
we
could
Kreger
Maas.
Can
you
comment
anything
about
the
support
for
the
nano
server
on
Windows
Server,
2000
19?
E
A
So
I
have
a
question:
I'm
assuming
we
do
have.
Some
images
are
non
nano,
sorry
right
as
part
of
having
comprehensive
test
coverage.
I'm.
Assuming
that
you
have
nano
server
for,
like
you
know,
let's
say:
90,
95
%,
there's
gonna
be
some
things:
I
have
Windows
server
core
or
even
full
windows,
because
we
may
have
some
upstart
testing
specific
things.
Do
we
have
that
in
our
test
matrix
or
we
can
switch
a
hundred
percent
to
nano
server
so.
C
F
This,
which
is
100%
honest
server
before
again,
it
was
the
post
image,
as
Mark
said,
and
there
were
what
a
couple
of
tests
which
were
using
specifically
nano
server
like
the
tests
which
were
using
the
resource
consumer
image,
since
there
was
a
mono
server,
another
one
which
was
falling
from
public
or
private
registry,
which
was
also
no
GCR,
which
is
another
server
image.
Well,.
A
Let
me
let
me
try
to
rephrase
that
I
think
I
messed
up
a
lately
the
question
and
so
are
adding
support
for
container
deriding
or
moving
away
from
dock
regime.
One
of
the
important
things
of
actually
certifying
and
validating
the
windows
containers
work
well
on
kubernetes.
Is
that
to
test
all
flavors
of
windows
containers
at
Carano
kubernetes,
so
that
means
testing
windows,
server,
core
a
nano
server,
full
windows.
A
So
as
part
of
our
test
matrix,
we
should
have
test
cases
that
instantiate
containers
of
all
three
SKUs
that
or
all
three
editions
of
Microsoft
has
and
run
test
cases
across
them,
because
that's
going
to
be
the
minimum
bar
of
saying
that
as
a
customer
feel
free
to
run
any
windows
container
here
right.
So
maybe
we
do
that
as
part
of
the
test
cases
for
continuity,
because
that's
where
we're
making
the
first
major
change
on
the
container
runtime
I
see.
C
And
we've
for
week,
I
think
that
yeah,
that's
a
very
good
point.
Michael
I
think
we
could
add
some
tests
with
either
this
like
a
slow
label
for
testing
the
the
larger
images
and
if
we
can
have
all
the
images
that
are
targeted,
the
the
non
slow
labels
that
could
still
probably
give
a
lot
of
benefit
to
the
test
passes
that
run
only
parallel
tests.
B
A
C
C
A
C
Yeah
I
wanted
to
see
how
much
interest
there
was
for
moving
or
and
like
what
it
would
take
to
say
that
other
versions
of
Windows,
like
sack
releases
of
Windows,
were
supported
for
kubernetes.
I
was
wondering
if
anybody
had
any
kind
of
interest
in
pursuing
this,
and
also
seeing
if
we
could
start
a
little
bit
of
get
a
little
bit
of
a
definition
of
what
it
means
to
support
these
different
operating
systems.
C
A
So
so
right
now
we
support
several
19
2009
teen
and
1903
right.
That's
the
the
two
ones
that
we
support
today.
I
think
that
the
the
barrier
for
before
when
Patrick
and
I
were
talking
about
it
in
variety
of
meanings,
was
all
around.
You
know,
effort
and
time
and
resources
right.
So
you
know,
given
the
windows
is
backwards
compatible
and
both
the
features
and
functionality
forward
is
always
improving.
I
think
it
makes
sense
for
us
to
support
them,
but
I
will
probably
say
you
know
this
is
my
personal
opinion.
A
Only
supporting
like
three
OSS
at
all
times
should
be
more
than
enough
like
we
can
support
the
latest
LTS
release,
so
Server
2019
and
then
maybe
two
of
the
SAC
releases
I
think
that
potentially,
if
we're
trying,
if
you
have
unlimited
bandwidth
and
time,
obviously
we
can
support
more,
but
I
think
that
those
those
two
SAC
releases
should
be
more
than
enough
and
then
the
LTS.
But
what
do
other
people
think.
B
A
A
B
C
C
G
A
So
I
mean
essentially,
we
could
do
something
like
that.
You
know
so,
if
you
have
let's
say
three
or
four
releases
that
we
support
every
day,
we
just
run
a
different
one.
I
mean
I.
Don't
see
a
reason
why
this
wouldn't
work
like
every
day.
You
run
a
test,
you
just
change.
What
what
you're?
What
you're
testing.
C
D
I'll,
just
I'll
chime
in
I've
I
think
the
ideal
case
is
that
every
new
to
me,
what
seems
ideal
is
that
every
new
Grenada
is
released
every
quarter
we
tested
on
2019
the
long
term,
long
term
image
and
the
latest
si
si
image.
So
there's
sort
of
a
one-to-one
correspondence
between
kubernetes
versions
and
si
si
releases
I
understand
that
the
timing
and
the
you
know,
that's
a
lot
of
consistent.
Regular
updates
to
the
test
runs
that
we
do,
but
if
we
can
achieve
that,
it
feels
sort
of
straightforward
know.
What
well.
A
Yeah
I
mean
for
public
cloud
providers,
I
think
that
easier,
I
think
for
enterprises.
It
takes
a
quite
a
while,
even
though
you
have
the
semi-annual
channel
that
duties
and
you
know-
and
in
you
know,
the
support
is
only
for
a
few
months
still
focused
on
update
right
away.
I
mean
there's
a
significant
effort
for
an
enterprise
to
update
their
OSS.
So
I
think
that
you
can
really
support
the
latest
because
I'll
give
an
example.
A
So
so
so
when,
when
when
that
comes
out
with
only
one
two
month
gap,
it
might
not
be
enough
time
to
basically
tell
people
who
only
support
that
and
Windows
Server
2000
19,
so
I
guess
long.
Runways
definitely
support
the
latest,
and
maybe
you
can
support
the
one
before
that
too.
So
one
or
two,
ladies,
like
what
Mark
wrote
there
and
if
the
effort
in
terms
of
resources
and
time
is,
is
basically
adjust
such
that
if
we
run
a
test
pass
every
day
and
every
day
we
said
we,
it
has
a
different.
C
E
A
E
I
was
about
to
say
that
I
I
think
one
of
the
important
things
coming
out
of
it.
It
would
be
if
we
can
clearly
document
something
about
I
think
from
LTS.
The
point
of
view
is
very
clear
that
the
2019
is
the
only
one
we
support,
but
from
sac
perspective
like
what
exactly
we
support
and
where
are
the
gaps
or
what
are
the
known
bugs
if
something
like
that
could
be
published,
but
I
think
we
get
a
lot
of
questions
around
that
right.
Yeah.
A
C
E
A
A
C
A
A
Needs
to
get
overhauled
yeah
if
someone
has
resources
and
can
take
a
look,
I
mean
all
of
the
things
are
still
right.
You
know
some
of
the
functionality.
You
know
the
IDs
continually
stuff
that
the
mark
updated
a
lot
of
functionality.
I
know
that
when
we
get
CSI,
probably
gonna
come
and
update
more
of
this
work
that
that
that
deep
had
added
earlier.
If
someone
has
resources
and
can
take
a
look
at
and
kind
of
like
a
couple
of
areas
that
you
can
update,
you
can
surgically
go
and
do
that.
Okay,.
E
C
E
A
I
mean
it
might
be
worthwhile
to
to
break
it
down.
I
think,
like
storage,
networking
and
compute
could
have
their
own
sections
and
me
and
Craig
kind
of
wrote
this
all
three
pages
back
at
the
beginning.
This
cuz,
we
didn't,
have
a
talk
right
there
and
we
had
to
get
something
out.
So
we
put
everything
into
three
pages
and
we
called
it
general
support
the
wind,
node
support
and
then
schedule
in
container
cell,
but
yeah
I
mean
if
someone
has
a
bandwidth
that
can
take
a
little
cooking,
so
Michael
I.
A
E
C
E
A
I
mean
the
biggest
part
of
this
page
was
API
right.
That
was
really
where
we
had.
You
know
when
folks,
when
I
tried
to
use
X,
Y
or
Z
on
on
Windows.
What's
there
and
what's
not,
there
and
I
have
a
feeling
that
once
container
D
support
comes
in
the
gap
between
was
available
for
Linux
versus
windows
should
shrink
mark.
Is
that
pretty
much
accurate
yeah?
It.
C
Should
there's
a
number
of
things
that
are
that
we
think
we
can
support,
continue
D,
that
Gertler
graceful
termination
period
on
single
file
mounts
which
was
I?
Think
another
big
one
where
there
was
a
couple
of
issues
and
people
I've
been
trying
to
keep
another
board
in
the
kubernetes
org
that
Patrick
set
up
up
to
date.
I
can
drop
that
link
in
the
chat
to
in
a
minute
and
I.
Think
that
had
a
lot
of.
C
F
C
H
I'm,
the
I'm
leading
the
effort
for
getting
Windows
workloads,
enabled
and
shift
on
at
Red
Hat,
and
what
we're
trying
to
do
is
also
contribute
to
upstream
so
Sebastian
on,
my
team
has
been
fixing
unit
tests
and
things
like
that
in
the
last
week,
or
so
so,
if
there's
any
more
like
low-hanging
fruit,
that
people
can,
you
know,
start
work
on
were
kind
of
fresh
to
coop.
To
this
whole
new
windows,
world
it'll,
be
helpful.
C
E
Right
and
and
I
think
in
the
backlog
mark.
There
are
a
lot
of
good
first
entry
tab,
whatever
it
is
called
the
label
good
first
or
something
like
that,
yeah
good,
for
something
like
that
yeah
yeah
there
are
actually
already.
There
are
a
lot
in
the
backlog
which
say
good
first
issue
and
it
probably
may
be
a
good
one
to
get
it
started.
I,
don't
know
if
it's
all
up
to
date,
but
you
know
there
are
a
few
here
and
there
that's.
H
E
A
And
obviously,
by
the
way,
since
we're
on
the
topic
of
darks
earlier,
as
you
guys
are,
either
ramping
up
your
knowledge
and
windows
and
kubernetes,
and
you
see
something
that
doesn't
align
you
know.
Usually
a
new
member
to
the
team
is
the
best
ones
to
actually
tell
us
that
hey
I've
read
this,
but
this
is
not
what
what
it's
in
reality
are.
You
know
are
great
kind
of
testers
of
the
dogs
and
helping
us
kind
of
update
them
or
at
least
flag.
There
is
a
need,
update,
okay,.
I
Absolutely
absolutely
I'm,
looking
into
like
a
beaut
windows,
continue
image,
unlike
Linux
machine,
so
I
tried
is
a
lot
I
think
it's
mostly
working
and
I
just
got
to
know
doc.
I
also
have
a
beaut
ex-con
ptex.
Actually,
it's
also
allow
you
to
build
container
windows,
cadena
image,
linux
node
from
there
of
official
document.
It
mentioned
you
taxes
for
like
multi-platform.
It
lists
like
a
linux
arm,
linux
AMD,
but
it
doesn't
mention
windows
and
I'm
just
wondering
if
anyone
use
ever
use
dispute
axe
before
you
know.
I
do
think.