►
From YouTube: Kubernetes SIG Windows 20230919
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
Hey
folks
welcome
to
the
September
19
Sig
Windows
community
meeting.
The
meeting
is
recorded
and
uploaded
to
YouTube.
So
please
add
here
to
the
cncf
code
of
conduct,
which
basically
translates
to
be
nice
and
kind
to
everyone
all
right,
so
announcements,
the
there's
one
week
left
now
for
the
steering
elections.
So
all
folks
who
are
on
the
call
or
insect
Windows
who
are
who
have
been
tagged
for
you,
know
being
able
to
vote.
Please
go
ahead
and
vote.
A
B
Nope,
that
was
that
was
the
big
one,
I
think
historically,
the
Syrian
committee
voting
is
the
turnout
hasn't
been
as
high
as
people
would
like,
and
it
doesn't
take
that
long.
So,
if
you
have
a
minute,
please
just
go
ahead
and
do
that
it's
much
easier
now,
with
a
new
platform
too,
just
you
just
sign
in
off
with
GitHub,
and
then
you
get
to
go.
A
Yeah,
it
took
me
maybe
five
minutes,
so
it's
not
it's
not
hard
at
all,
including.
B
Contributors
129
release
cycle
is
underway.
We
should
probably
remind
everybody
what
the
enhancement
freeze
data
is,
which
I.
Let
me
look
that
up
real.
C
B
It
looks
like
enhancements.
Freeze
is
October
5th.
A
A
C
A
But
if
there's
anyone
who's
new
I'll
give
a
couple
of
seconds
for
folks
to
speak
up.
If
not,
we
can
move
on
to
the
agenda.
A
All
right,
let's
move
on
to
the
agenda.
The
main
item
on
the
agenda
is
kirtana's
image
pool
per
runtime
class
cap,
which
is
not.
This
is
the
chase
issue.
A
So
I
don't
know
how
many
folks
on
the
call
are
familiar
with
this
kept
basically
has
to
do
with
supporting
image
pulls
per
runtime
class.
This
will
basically
help
Windows
a
lot
in
some
sense.
If
you
have
a
Windows,
node
and
you're
trying
to
bring
up
hyper-v
isolated
containers
on
them,
you
are
host
image.
You're.
Sorry,
The,
Container
image
doesn't
need
to
match
the
host
image
at
that
point,
so
you
could
have
a
2019
image
running
on
a
2022
host
and
I.
Think
for
to
make
that
happen.
A
What
we
require,
what
kirtana
is
proposing?
Is
you
also
when
you
ask
for
the
pool?
You
also
mention
the
runtime
class
as
possible
as
part
of
the
requester
to
the
runtime,
so
we
can.
How
do
you
folks
want
to
handle
this?
Do
you
want
to
go
through
this,
or
do
you
just
want
this
because
I
started
reading
it?
It's
pretty
it's
a
pretty
large
cap,
I
started
giving
some
comments.
So
how
do
you
folks
want
to
tackle
doing?
This
is
Keystone
on
the
call?
A
A
So
so
Mark,
why
don't
we
give
this
as
a
sort
of
an
announcement
so
that
at
least
folks
on
the
call
are
ever
and
I
I
know
James
posted
about
this
on
on
slack,
so
I
guess
another
reminder
for
folks
to
start
reading
the
cap
and
start
providing
feedback.
I
I
in
fact
started
doing
it,
but
it's
pretty
large.
So
it's
not
something
you
can
do
in
five.
Ten
minutes
you
need
to
put
aside
at
least
in
r
or
so
I
think
to
read
then
digest
the
cap
yeah.
B
And
I
think
with
this
one
too,
like
it's
a
lot,
there's
a
lot
more
complication
like
it's
a
lot
more
complicated
than
it
seems
initially,
because
it
goes
into
scenarios
like
garbage
collection
like
what
happens
if
you're
using
CRI
CTL,
and
things
like
that
like
like
what
are
all
the
implications
for
that
just
doing.
The
initial
poll
is
fairly
straightforward,
but
there's
a
lot
of
kind
of
second
order:
side
effects,
yeah.
A
Yeah
I
was
thinking
Mark
like
we
could
even
try
and
remove
the
complications
around
CRI
and
CRI
CDL
from
this
cap
at
least.
Otherwise
it's
going
to
get
very
noisy,
so
we
should
try
and
keep
this
I
I
I'm
thinking.
Maybe
we
keep
the
boundary
around
the
cubelet,
the
runtime
and
what's
needed
in
that
area.
If
we
start
to
go
to
the
tooling
side,
it's
going
to
get
more
complex
right.
B
I
I,
it
depends
on
the
solution
like
I
think.
The
solution
here
is
like
one
of
the
proposals
is
to
have
everything
plumbed
through
the
CRI
API
and
depending
on
that,
like
that,
could
make
it
not
as
complicated
but
yeah
I.
Think
it'd
be
good
to
try
and
separate
that
out
or
but
probably
still
call
out.
C
That's
a
great
question:
I
was
thinking
about
when
I
was
reviewing
it
yesterday
and
I
forgot
to
ask
it
so
because
that's
I
think
cryo
would
we'd
want
to
make
sure.
But
although
I
don't
know,
if
cryo
has
Windows
or
does
it
it
doesn't,
but
this
wasn't.
B
There
was
another
meeting
like
this
was
discussed
either
this
or
a
very
similar
PR
at
the
same
time
was
discussed
in
signode
and
Peter
hunt
was
was
interested
but
I.
Think
I,
don't
remember
if
cryo
already
kind
of
did
some
of
this
I
think
he
was
more
like
keep
me
updated
or
they're
more,
like
keep
me
updated
and
then
we'll
see
what
happens
after
that.
A
Yeah
and
Kingston
also
presented
the
cap
at
the
image
working
group
that
they
have
that's
under
signode
yeah,
so
she
presented
that
I,
don't
remember
what
they
spoke
about
with
respect
to
cryo
there
at
the
top
of
my
mind.
So,
but
it's
a
question
that
either
I
I
can
post
on
the
cap
as
part
of
reviewing
this
yeah,
because
it
can't
be
just
continuity
specific.
It
has
to
also
support
so
maybe.
A
Right,
if
folks
have
nothing
further,
this
is
just
in
more
like
a
hey.
There's,
a
cap
out
there
that
we
all
need
to
review
sort
of
announcement
and
then
we
can
try
and
figure
out.
If
we
can,
you
know
maybe
have
kids
are
not
come
in
once.
We've
reviewed
it
and
do
a
follow-up
with
her
just
to
keep
this
going
because
October
5th
is
and
or
October
6th
is
going
to
come
up
pretty
soon,
and
we
should
try
and
get
this
in
for
129.
A
All
right,
let's
move
to
the
next
agent
item,
which
is
about
how
to
set
up
proxy
for
Windows
or
continuity.
I
was
actually
hoping
that
Monty
would
be
in
this
call,
but
I'm
not
sure
she's
attending.
A
So
the
problem
in
this
in
this
issue
is
proxy,
has
been
set
up
on
the
Windows
node,
but
continuity
is
not
using
the
proxy
to
pull
images.
A
However,
we
just
implemented
proxy
support
in
openshift
for
Windows
nodes
and
we
did
get
it
working.
So
that's
why
I'm
just
hoping
Montes
here
who
actually
worked
on
it
and
she
might
be
able
to
help.
But
I
don't
see
her
here.
A
A
A
A
A
So
Jameson
mark
on
Azure
there's
no
proxy
support
or
if
that's
not
a
feature,
that's
had
been
added
yet.
B
I
think
that
the
some
of
the
recommendations
that
at
least
Azure
gives
is
to
mirror
content
into
an
Azure
container
registry
and
pull
from
that.
So
I
don't
know
if
it's
something
that
people
are
like
I,
at
least,
if
you're
there,
following
all
the
Azure
documentation,
they're
more
likely
to
go
that
route
than
trying
to
do
practicing
for
the
container
registries.
A
If
they
were
able
to
solve
the
problem,
because
that's
the
last
thing
I
see,
Monty
has
asked
a
person
who
opened
the
same
issue.
I
think
it's
the
same
yeah,
it's
the
same.
Bruce,
that's
on
the
continuity
Slack.
A
The
last
question
Monty
is
us
is
if
they
have
tried
setting
the
proxy
environment
variable
at
the
Machine
level,
so
that
it
can
be
inherited
by
the
continuity
process
instead
of
configuring
it
from
the
continuity
proc
config,
and
that
seems
to
work
for
us,
but
I
would
like
to
see
if
they
have
tried
it
and
did
that
not
work.
Because
then
you
know
we
need
to
look
into
this
further.
C
A
A
Yeah
except
require
that
notary,
but
that
was
a
bit
of
a
pain,
but
so
we
had
to
come
up
with
a
way
for
the
operator
to
request
to
sort
of
coordinate
reboots
right,
because
you
can't
just
can't
reboot
all
the
nodes.
At
the
same
time,
you'll
have
services
or
you'll,
you
could
potentially
have
workloads
running
and
they
would
go
down
so
you
actually
have
to
drain
it.
A
B
D
D
So
question
I
know
you
just
you
just
did
talk
about
handling
reboots
and
I
was
looking
into
this
for
something
else,
but
are
there
any
other
sort
of
like
best
practices
to
consider?
If
you
need
to
kind
of
set
up
infrariate
things
you
need,
do
you
require
reboots.
A
B
Yeah
I
think
probably
the
best
practice
is
to
try
and
do
the
reboot
before
you
flip.
The
node
directly
in
podcast,
scheduled
I,
know
that's
not
always
possible,
but
that's
probably
the
safest
way
to
do
it.
A
Yeah
I
I
think
it
becomes
tricky,
especially
in
the
case
of
like
proxy.
You
could
set
up
a
proxy
after
the
node
has
been
configured
and
after
you've
had
workloads
running,
and
then
you
want
to
inject
proxy
configuration.
That's
when
it
becomes
tricky
where
you
have
to
make
sure
that
the
node
is
drained
and
you're,
not.
You
know,
taking
down
workloads
because
you've
decided
to
reboot.
After
applying
the
proxy
configuration.
A
A
These
are
more.
If
you
have.
You
know
like
workloads
that
you
you've
added
a
deployment
to
the
node.
Those
things
would
need
to
get
drained,
but
things
like
our
that
are
Damon
said:
there's
no
draining
them
right.
You
just
want
them
running
on
every
on
every
note.
That's
out
there,
so
they'll
come
back
up
the
next
time.
The
node
reboots.
B
A
Yeah,
that's
a
good!
That's
a
good
point
mark.
So
maybe
you
could
use
this.
You
know
the
seven
minutes
or
whatever
we're
giving
back
to
you
folks
too,
to
encourage
you
folks
to
vote
for
the
scaring
elections.