►
From YouTube: Kubernetes SIG Windows 20200324
Description
Kubernetes SIG Windows 20200324
A
Hello,
everybody
and
welcome
to
another
sick
windows
meeting
it's
the
24th
of
March
and,
as
always,
is
a
recorded
meeting.
So
please
adhere
to
the
CN
CF
code
of
conduct.
We
had
a
pretty
busy
last
couple
of
weeks.
I
think
that
you
know
the
between
the
PRS
that
one
in
and
you
know
great
job
by
everybody-
the
the
code
changes
I,
wouldn't
in
great
great
work
by
everybody
as
well.
You
know
the
broad
windows
to
be
in
a
very
good
spot
for
for
118.
A
There
was
one
late-breaking
change
that
came
in
last
week
and
it
was
around
some
test
failures
that
were
happening,
but
that
was
very
quickly
resolved,
so
appreciate
the
work
that
the
Lynne
and
her
team
and
others
have
been
doing
to
kind
of
resolve
those.
So
let's
go
ahead
and
look
at
our
agenda,
so
I'm
gonna
share
my
screen
as
well.
That
way,
we
can
all
see
what
do
you
have
here
so
the
first
one
mark
1:18
test
signal,
yeah.
B
So
I've
been
invited
to
the
release
meeting
twice
in
the
past
week
and
both
times
I've
been
asked
about
this.
So
I
just
want
to
raise
awareness
here
and
get
some
clarification
on
the
GCE
test
issue.
If
there's
folks
who
can
answer
that
on
the
call,
but
so
there's
still
too
flaky
test
bugs
for
1:18
the
first
one
is
about
this
CPU
resources
test
which
is
running
in
the
AKS
engine
test
passes,
and
for
that
one
I
believe
we
just
need
to
mark
the
test
case
of
serial.
B
B
But
the
release
team
is
concerned
that
that
GCE
Windows
1909
test
pass
is
running
so
few
of
the
test
cases
compared
to
all
of
the
other
test
cases
and
was
asking
if
we
should
keep
that
as
release
informing
or
not
because
they
thought
it
would
lead
to
more
confusion.
I
think
it
started
off
as
a
as
a
test
failures
which
Ben
and
you
and
Peter
took
a
look
at
yeah.
A
So
I
don't
have
much
context
on
that,
I
mean
so
so
I
understand
there
should
it
be
released,
informing
like
we
don't
support
1909,
yet
right
so
so
from
that
perspective,
it's
this
is
I,
don't
want
to
call
it
experimental
and
and
I
don't
see
Peter
on
so
so
so
everything
gives
us
a
signal
right.
So
anyway
of
you,
it
is
like
everything
has
information,
whether
it's
failing
or
not
it's
important
for
us,
because
then
it
means
we
need
to
pay
some
attention
to
it.
A
Should
it
be
released,
informing
I'll,
probably
say
no
for
two
reasons:
it
doesn't
run
the
full
test
to
it,
like
you
said,
and
it's
on
a
build
of
Windows
that
we
don't
fully
support
yet
right,
because
you
haven't
pervasively,
run
tests
across
the
board
with
it.
How
do
you
kind
of
agree
with
that
and
only
hear
Ben's
opinion
as
well?
I.
C
Mean
yeah,
but
I
think
the
problem
here
might
be
like
it
seems
like
there's
just
something
wrong
with
these
GC
tests,
or
even
there
2019
tests
are
running
like
not
the
entire
suite
I'm,
not
sure.
What's
going
on
there,
it's
not
like
I
mean
there
are
red
test
tube,
but
it's
like
this.
When
you
look
at
the
test
results
they're,
like
super
sparse,
you
know
like
it.
One
run
runs
like
a
certain
number
of
tests
on
the
next
run
runs
like
a
different
number
of
test
line.
I,
don't
know.
What's
going
on
there.
A
C
B
C
D
B
B
B
C
A
B
A
B
A
E
E
We
need
to
update
for
a
caius
means
our
data
hosts
to
February
or
March
in
this
Mark's,
because
the
image
is
already
out,
but
for
that
we
need
to
also
update
the
ete
test
images
that
are
in
the
docker
hub
repo
that
we
share
with
Google
with
Google
and
with
the
Cuba
DMS,
and
we
need
a
coordinated
effort
here.
So
if
we
were
to
move
to
March
right
now
and
thus
update
the
container
images
as
well,
that
would
mean
that
we
will
break
Google
tests
because
they
don't
have
the
the
proper
host
image
version.
E
I
think
I'm,
not
sure,
then
I
was
hoping
people
or
somebody
was
here
so
I.
So
we
we
know
what
hosts
they're
using,
but
the
images
are
using
for
the
hosts.
While
we
propose
and
I've
discussed
this
with
with
closure
and
then
with
Mark,
is
that
we
move
the
container
images,
the
January
ones
right,
the
ones
that
we're
using
right
now
to
a
separate
repo,
each
we
legacy
or
oh,
it's
called.
E
And
this
way
we
can
move
a
case
to
the
March
host
image
and
thus
push
the
container
images
as
well.
And
this
way
Google,
who
is
depending
on
our
images,
can
still
use
the
old
ones
until
they
want
to
update
their
hosts.
A
E
F
A
As
well
as
potentially
Cuba
DM
and
let
Google
upgrade
their
cadence
I'm,
okay
with
that
give
Peter,
also
a
buzz
on
on
slack,
maybe
create
a
private
conversation
with
him.
If
it.
If
they
are
already
on
the
past
February
patch,
then
they
might
be
okay
right
I
mean
the
issue
happen,
either
before
February
or
after
February.
So
if
you're
after
February
you're
good
so
so.
F
E
So
we'll
we'll
be
in
contact
with
them,
but
of
course
we
should
buy
and
the
base
okay,
baby
em
works
as
well.
Also
I
have
a
PR
that
updates
the
host.
E
A
G
G
Thank
you,
so,
okay,
so
mark
and
I
have
been
talking
about
it
and
I've
been
I.
Have
a
couple
of
things
to
discuss
here,
but
the
first
thing
I
want
to
start
and
some
of
the
I'll
discuss
in
the
in
the
future
meetings
as
well,
but
first
of
all
the
first
thing
I
want
to
discuss
is
this:
is
the
backlog,
sorry,
the
board,
created
by
Patrick
a
while
ago,
I
find
it
useful
and
I
wanted
to
talk
about?
G
A
May
I
can
talk
a
little
bit
about
why
you
know.
Patrick
and
I
still
use
this
and
using
maybe
once
every
couple
of
months
to
kind
of
track
our
release
and
look
where
we
are
and
kind
of
look
at
the
prioritization,
but
we
used
it
the
most
extensively.
When
were
tracking
114
worried
there
was
a
specific
number
of
items.
I
will
have
to
finish
to
call
ga
where
it
was
paramount
that
they
actually
finished
them
all
after
1.14.
A
Now
you
know
if
things
are
ready
and
they're
done
for
a
specific
milestone,
they
Kadeem
and
if
not,
they
don't,
and
then
we
track
bugs
on
as
a
one-off
in
our
meeting.
So
that's
why
we
haven't
explicitly
used
the
board,
what
we
show
it
to
everybody
on
on
a
on
a
weekly
basis,
but
I'm
not
opposed
to
spending
a
few
minutes
every
week.
A
If
we
want
to
talk
about
specific
issues
on
the
board
and
how
we
prioritize
them,
I
just
I,
don't
think,
there's
enough
of
that
to
basically
justify
more
time
on
the
board
like
we
can
do
some
of
the
work
on
the
back
end
in
private,
like
we
did
with
Patrick
and
to
help
prioritize
and
put
things
at
the
spotlight
rather
than
having
everybody
look
at
it,
but
yeah
we're
always
open
to
suggestions
if
something
works.
Better.
Okay,.
G
Yeah,
because
yeah
that's
what
I
was
thinking
as
well
and
that's
a
fair
point:
I
think
you
know
if
I
mean
I,
think
it's
better
to
use
it
a
little
bit
at
least
every
other
week
or
every
week
to
just
get
a
better
sense
of
where
we
are
so
just
based
on
that
mark
and
I
and
even
Patrick
before
we
left.
We
we
thought
about
how
we
can
organize
these
votes
a
little
bit
better
as
well.
G
So
I
created
these
two
boards
on
my
private,
not
private
or
public
repo,
but
we
can
move
them
to
the
main
repo.
But
I
don't
know
if
this
is
gonna
be
useful.
So
what
I
did
was
I
went
to
the
whole
board
and
I
categorize
them
based
on
prioritize
so
sigmund
those
prioritized
and
whichever
items
had
priority,
I
put
them
in
the
bucket,
so
I
have
the
the
Sikh
community,
the
kubernetes
all
up
community.
G
You
know
have
these
three
priorities
so
priority
critical
origin
for
a
different
zone
and
priorities
long-term,
and
these
are
the
priorities
currently
assigned
to
these
issues.
So
it
kind
of
we
get.
We
can
look
at
what
issues
are.
You
know
from
priority
point
of
view,
and
then
we
also
have
awaiting
more
evidence
or
no
priority.
The
other
board
I
created
was
based
on
the
category.
So
this
is
more
of
bug
versus
feature
versus
failing
test
versus
API
change.
I
know
there
are
other
areas
as
well.
G
G
A
You
know
the
the
drive
features
like
based
on
the
priorities
that
the
sig
said
and
looks
at,
and
then
people
pick
opportunistic
bugs.
One
I
think
one
thing
that
you
might
want
to
put
in
your
thing,
Moz
that
might
be
very
useful,
is
a
good
first
issue
or
entry-level
issues.
Think
about
something
along
those
lines.
So
if.
A
New
to
the
sig
and
they're
looking
for
something
easy
to
fix
to
just
gain
confidence
and
and
start
testing
the
waters
tag
some
of
those
bugs,
because
those
are
the
ones
are
easier
that
don't
require
you
to
add
new
test
cases.
They
it's
a
simple
fix,
simple
to
verify.
If
you
can
tag
those,
that's
the
only
addition
to
all
the
work
that
you've
done
so
far.
That
I
would
say:
okay,.
G
A
G
G
Thing
Michael
related
to
this
prioritization
is
I,
see
a
lot
of
items.
Do
not
have
a
priority
currently
and
I
just
wanted
to
understand,
like
maybe
that's
the
same
meeting
where
we
start,
you
know
prioritizing
stuff,
it
could
be
just
better
to
have
a
priority
on
on
stuff
so
that
we
don't
you
don't
miss
any
any
particular
stuff.
All
right,
I
agree,
okay,
so
so
that
was
it
from
a
board
point
of
view.
Thank
you
for
the
feedback.
G
If
there
is
any
feedback,
please
let
me
know
on
slack
so
the
next
item
I
am
working
on
is
basically
I
think
it
goes
to
your
higher
level
priority
for
119,
so
basically
I'm
creating
a
markdown
file.
It
probably
would
be
very
next
week
we
are
going
through
all
the
tips.
So
if
you
look
at
all
the
sorry
ka
PS
and
what
is
required
for
what
a
19,
for
example,
going
into
the
going
into
the
graduation
criteria
for
different
enhancements
that
are
coming
in
and
then
kind
of
making
like
okay.
G
So
this
is
what
looks
like
for
119,
and
then
this
is
gonna.
What
looks
like
for
120
kind
of
making
more?
Like
a
roadmap
kind
of
thing
for
the
community,
because
I
kind
of
have
heard
this
feedback
very
often
and
me
personally,
when
I
look
at
it,
I
kind
of
get
confused
on
what
exactly
is
coming
for
119,
especially
a
new
person,
so
I
will
try
to
create
some
kind
of
document,
and
then
we
can
convert
to
a
markdown
file
and
post
it
here,
depending
on
the
feedback
next
time.
G
G
A
So
far
our
prioritization
was
very
ad
hoc
because
it
was
based
on
need
were
always
behind
8-ball,
so
we're
always
kind
of
chasing
the
the
biggest
crush
was
the
one
that
got
priority.
You
never
had
time
to
say:
hey.
We
have
resources
to
spend
and,
let's
probably
prioritize
things,
that's
just
how
the
reality
was,
for
this
sake,
right.
A
Absolutely
and
we
appreciate
that
cool,
so,
let's
we
have
a
few
more
minutes
left,
so
I
want
to
talk
a
little
bit
around.
You
know,
priorities
for
for
the
nest,
release
and
kind
of
start
identifying
some
of
the
people
that
cannot
start
driving
some
of
that
work
and
then
figure
out.
You
know
what
are
some
of
our
next
at
the
high
level,
so
the
first
one
is
container
D
mark.
Is
it
safe
to
assume
that
you
will
be
kind
of
leading
that
effort?
A
B
Right
now
it
might
be,
but
that
will
I
think
the
next
thing
that
we
really
need
to
do
in
order
to
get
more
clarity,
and
if
it's
feasible,
to
go
to
19
is
to
start
getting
some
test
passes.
Some
more
test
passes
with
running
container,
d2
and
and
windows
to
see
kind
of
what
the
the
feature
gaps
are
between
the
docker
shipment
container
D
got.
G
G
To
add
on
I
didn't
get
a
chance
to
work
with
Jeremy.
He
was
busy
and
I
was
busy
as
well,
but
this
week
I'm
looking
to
work
on
G,
MSA
and
UNIX,
the
male
socket
mouse,
which
is
needed
for
container
D
graduation
to
beta.
What
we
are
trying
to
understand
is
what
is
needed
in
the
you
know
in
the
karai
container
D
at
their
platform
level
and
what
is
needed
from
a
kubernetes
perspective.
G
A
Was
pretty
busy
with
his
work
last
week?
So
that's
why
he
has
been
as
around
on
on
sick
windows,
but
he
said
he's
a
big
volunteer
on
our
sake,
so
yeah,
alright
cool.
So
is
it
safe
to
say
mark
that
for
container
D?
Our
goal
is
that
you
know
the
nones
bed.
Unless
you
have
parity
101
on
features,
maybe
stability
won't
be
there?
Maybe
some
bugs
will
be
there,
but
functionally
dr.
Schuman
continuity
should
be
the
same
for
Windows
is
that
is
that?
Is
it
a
good,
safe
yeah?
B
A
C
I,
don't
have
like
I
am
I'm
kind
of
a
new
team
right
now
at
VMware,
where
I
mean
I'm
working
on
cluster
API,
but
like
we're
driven,
not
it's
just
like
a
Windows
specific
thing.
So
it's
not
as
straightforward
like
I'm
working
on
cluster
API
issues,
but
Windows
support
isn't
the
priority
right
now,
God.
So.
F
A
Cool
so
Anna,
specifically
when
I
talk
about
cluster
EP
I
am
talking
about
two
things.
There
was
some
initial
work
that
both
bended
and
swallows
Adelina.
You
have
a
couple
of
pages
I've
been
running
that
talk
about
how
you
can
take
Windows
nodes,
boostrap
them
through
the
through
queue
medium
for
cluster
API,
on
the
different
public
cloud
providers
right
and
and
I
know
we
have
when
I,
say
public
cloud
providers.
I
should
be
a
little
bit
more
open
here,
so
yws
as
your
hyper-v
and
VMware.
A
H
So,
basically
for
Jocelyn
the
perspective
work
right
now.
It's
not
gonna
be
upstream,
like
it's
not
any
copy
work.
It's
gonna
be
about
the
bootstrap
provider.
Kk
is
not
going
to
be
involved
in
this
project,
she's.
Actually,
switching
teams
and
I'll
be
there
more
an
advisory
role
just
like
with,
but
not
actively
developing
it
because
again,
like
because
of
priorities,
because
I'll
be
working
on
the
container,
D
and
hyper
ISA.
So.
A
E
I
hear
so
on
closer
API,
as
I
said:
I've
written
some
stuff
and
I'm
working
on
it
on
my
own
right
now,
I'm
making
a
lot
of
time
to
it,
but
just
to
to
see
how
I
can
use
it
in
in
some
other
situations.
For
my
for
my
use
and
one
I
have
that
I
wanted
to
start
rewriting
the
dock,
for
it
so
publish
it
on
what
we
already
have
and
I
wanted
to
start
writing
and
contribute
that
contributing
actually
to
that
project
to
build
images
that
support
cluster
API.
A
Okay,
so
that's
great
Adelina
there.
I
guess
what
you
might
be
looking
for,
though,
is
someone
also
to
come
in
and
help
you,
because
it's
gonna
be
probably
significant
development
or
or
integration
that
might
need
to
happen
to
enable
some
of
the
scenarios
you're
talking
about
so
love
the
fact
that
you're
working
on
this
on
your
own
time?
This
is
super
appreciate
that
these
people
can
get
you
some
help
as
well
and
and
which,
which
cluster
API
provider.
Are
you
looking
at
for
your
own
work?
If
there
are
you
concentrating
on
of
them,.