►
From YouTube: Kubernetes SIG Windows 20230926
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
welcome
to
the
September
26th
Sig
Windows
community
meeting.
This
meeting
is
recorded
and
uploaded
to
YouTube.
So
please
adhere
to
the
cnco
code
of
conduct,
which
basically
translates
to
be
nice
to
everyone
all
right,
so
any
announcements
other
than
just
reiterating
that
enhancement,
free
state
is
October
6th
and
we
are
tracking
one
enhancement
from
the
Windows
site.
A
A
All
right
folks,
another
reminder:
I'm
I'm,
glad
I've,
already
finished
voting.
So,
okay,
any
new
contributors
on
the
call.
A
Don't
see
anyone
that
I
haven't
seen
before
so
we
can
move
on
to
the
topic.
So
the
first
agenda
item
is
I.
Did
we
did
get
pinged
on
slack
about
if
we
are
targeting
any
features,
removals
or
deprecations
for
the
129
release
and
the
main
one
I
know
is
imageable
for
runtime
class
before
which
we
have
kept
going?
Is
that
anything
else
that
I
have
missed.
B
I
think
that's
the
the
main
one
I
think
we
should
still
try
and
make
progress
on
the
Node
log
query.
But
I
don't
know
if
I
think
we
talked
to
everyone,
I,
don't
know
if
that's
gonna
make
it.
A
B
I
think
that
they
were
asking
for
just
any
enhancements
coming,
but
double
check
like
any
any
state
changes.
Oh.
A
Any
state
changes,
I,
see,
I,
see
yeah
I'm,
not
100,
sure
I've,
just
not
had
the
Cycles
to
refocus
on
that.
I
was
hoping
Fabian
and
there
was
someone
else
from
Red
Hat
we're
trying
to
fix
the
end-to-end
tests,
but
I
don't
know
what
the
status
of
that
is.
I'll,
try
and
find
out.
A
A
But
again
you
know
we
still
have
a
month.
So
it's
not
a
lot
of
work.
I
have
some
of
it.
Written
I
just
need
to
go
back
and,
and
do
that
again,
so
my
target
for
it
to
hit
go
from
alpha
to
Beta
is
have
it
run
in
some
end-to-end
test.
You
know
on
the
on
the
Windows
side
and
have
this
Cube
cuddle
plugin
available
for
folks
to
use.
What
do
you
think
Mark?
That's
what
we
had
I
think
that's
what
we
put
down
in
the
cap
also.
A
Yeah,
let's
try,
if
not,
we
should
I.
I
will
definitely
re
refocus
and
do
this
for
1
30
but
129
is
might
be
touch
and
go,
but
we'll
see
so
so
the
takeaway
is,
let's
not
put
it
down
until
we're
sure.
Is
that
what
you're
saying
or
should
we
just
put
it
down
and
then
say
it
didn't,
make
it
I.
A
B
I
think
we
also
should
try
and
pick
up
the
effort
to
get
the
vertical
pod
Auto
scaling
stabilized
for
Windows.
That
would
be
really
good
to
finish.
Tracking
I.
Don't
know
if
that's
going
to
be
included
as
part
of
the
the
big
cap,
but
we
should
definitely
prioritize
that
too.
For
this
release
and.
A
All
right,
I
have
these
three
items
set
down
here
and
I
will
respond
to
the
thread
on
on
slack
with
the
person
who
ping
this
anything
else.
Folks,
if
not,
we
can
move
to
the
next
topic.
This
is
a
topic
I
think
Jay
brought
along
and
I
think
finally
Jay's
here,
so
we
can
I.
Think
Monty
is
also
here
who
might
be
able
to
add
to
this
so
Jay.
Do
you
want
to
do
you
want
to
talk
towards
this.
A
It
was
all
over
the
place.
There
was
a
mailing
list.
There
was
a.
There
was
a
couple
of
GitHub
issues
that
are
threads
on
Sig
Windows.
There's
a
thread
on
the
the
cncf
continuity,
so
yeah,
it's
everywhere.
D
I,
don't
even
I,
don't
even
yeah
I,
don't
know.
I
was
just
making
sure
that
we
talked
about
it,
but
I
don't
have
I,
don't
have
anything
to
say
about
it.
I
haven't
looked
at
it.
A
E
Yeah
I
kind
of
commented
on
that
issue.
That's
pointed
in
the
document,
so
I
think
even
James
commented
on
that
and
yeah
that's
that's
kind
of
been
a
workflow
we're
setting
it
at
Machine
level
and
we
usually
need
a
restart
of
the
system
to
for
that
to
be
inherited
but
dreams.
You
mentioned
on
that
issue
that
there
are
some
side
effects,
so
I
just
wanted
to
ask
you.
Can
you
elaborate
on
that?
C
Oh,
just
it's
that
it's.
If
you
set
it
at
the
Machine
level,
then
it's
going
to
affect
all
processes,
and
that
was
what
one
of
the
person
who
opened
up
the
pr
was
complaining
about
this.
They
didn't
want
that
to
they,
wanted
it
to
be
per
process
proxy
I,
guess
I'm,
not
sure,
okay.
So
that's
what
that
was.
A
I
I
think
I'm
I've
made
a
note
of
that.
So
it
looks
like
the
folks
who
opened
that
issue.
They
want
it
done
on
a
per
process
level.
So
I
guess
they
want
it
done.
Cubelet
I,
don't
know
Cube
proxy
identity
and
not
far
across
the
system
and
I
think
we
went
with
the
at
least
the
wave.
You
have
seen
proxy
being
used
with
our
customers.
Is
they
wanted,
like
you,
know,
node
wide?
A
D
A
A
A
B
B
B
D
That,
if
it
was
a
host
process
container,
that's
an
environment
variable
to
whatever
that
host
process
container
is
right.
If
you,
if
you've,
got
a
Python
program
and
it's
running
on
a
host
process
container,
it's
just
a
environment
variable
to
tell
it
to
use
a
proxy
right.
That
would
be
program
specific.
E
A
We
don't
know
how
they've
set
up
the
global
proxy
and
I
think
somewhere,
which
is
why
this
thread
is
this
issue.
Is
there
in
multiple
places
right
somewhere?
They
had
mentioned
that
they
don't
want
to
do
it
at
a
system.
Level
is
what
I
understand
they're,
not
able
to
find
where
they
have
said
that,
but.
A
But
I
think
we
have
open
questions
for
them,
at
least
on
this
issue
and
we'll
we'll
we'll
see
what
they
do.
But
otherwise
this
looks
like
it
could
be
a
bug
in
continuity.
If
there
is
a
option
to
set
it
in
the
in
that
to
config.tomo.
D
D
A
All
right
that
sounds
good.
The
only
other
thing
I
had
is
last
time
we
had
spoken
about
the
imageable
part
of
runtime
class
kept,
and
we
had
us
folks
do
take
a
look
at
it.
I
think
I
reviewed
it
so
did
James
I,
think
James,
or
was
it
Mark.
A
James
reviewed
it,
oh,
hey,
I
can
see
you
hey.
F
So
a.
A
A
Worries,
do
you
want
to
add
anything
to
this?
We're
just
making
sure
that
this
has
enough
visibility
and
folks
are
looking
at
it,
because
we
do
want
to
see,
try
and
get
it
into
129.
F
Yeah
I
was
planning
to
just
bring
it
up
in,
like
the
continuity
like
community
meeting
too
just
to
get
more
ice
from
I
think
like
Derek
and
Mike
Brown
as
well.
Okay,
but
the
other
work
group
that
has
been
meeting
every
Wednesday,
I
discussed
it
there.
So
Sergey
was
also
there
I'm,
just
waiting
on
reviews
from
them.
A
Okay,
so
what
I
would
suggest
yeah,
I
I
think
I
mentioned
it
also
right,
I
think
the
to
get
this
kept
into
the
kubernetes
captain
yeah,
it's
better
that
you,
you
sort
of,
don't
mention
the
level
of
detail
you
mentioned
about
continuity.
There
keep
it
very
generic
as
far
as
what
you
need
a
runtime
to
do.
A
Excellent
yeah,
so
then,
let's
just
wait
for
the
notebooks
to
review
and
if
needed,
you
can
also
add
it
to
the
node
a
node
agenda
and
have
it
looked
at
there
to
just
so
that
we
have
the
kept
merged
by
the
deadline.
Okay,.
F
And
one
just
additional
question
is
how
do
I
sync
up
with
Clio
folks
like
once
this
cap
goes
through
I
work
on
the
changes
that
container
the
needs,
but
I
think
I
brought
it
up
in
the
work
group
meeting
as
well.
F
How
do
I
follow
it
up
and
make
sure
like
that
also
goes
in
exactly
around
the
time
that
we
also
make
continuity
changes,
because
only
then
I
think
it
would
make
sense
to
change
the
cubelet
right
yeah.
All.
C
A
The
cryo
folks
are
usually
at
Sig
note
and
I
can
help
with
that.
If
your
findings
go
downstairs,
I
know
both
Peter
and
Ronald
and
I.
Can
you
know
we
can
work
with
them
and
see
what
what's
needed,
but
the
main
thing
is,
of
course,
right
like
this:
whatever
change
you
make,
the
cubelet
should
be
able
to
work
with
the
with
with
cryo
that
doesn't
have
the
changes
so
good.
F
A
A
All
right,
we've
come
to
the
end
of
the
agenda.
Do
folks
have
anything
more
to
discuss.
D
I
I
suppose
we
have
a
small
announcement,
which
is
good
news
for
us,
which
is,
as
you
know,
we've
been
struggling
with
developer
environments
for
a
long
time
and
the
cncf
we
worked
with
test
infra
and
the
cncf,
and
at
least
for
the
stuff
we're
doing
right
now
with
trying
to
do
the
operating
and
stuff
and
the
win
service
proxy
stuff.
They
gave
us
eks
accounts
that
we
can
use
for
development
on
those
projects.
D
Basically,
so
if
somebody
feels
like
they
need
to
do
some
windows
testing
and
they
don't
have
a
cloud
account
and
they
want
to
sort
of
get
involved
with
Sig
Windows
like
reach
out
on
slack
start
a
thread,
and
you
know
we
can
work
with
r
naught
to
to
be
able
to
get
people
some
development
environments
for
testing
Windows
clusters
and
developing
features
and
stuff
obviously
like
if
you
have
an
account,
if
you
have
a
cloud
account,
if
you're
able
to
do
this
stuff,
like
it's,
not
we're
not
trying
to
waste
cncf
money.
D
But
you
know.
A
D
A
Operating
us
and
service
proxy
right,
yeah.
A
D
So
thanks
to
test
infra
and
arnod
and
dims
and
other
folks
who
approve
that
and
and
of
course,
to
Amazon
for
donating
that
money
to
the
cncf,
because
I
think
that's
how
that
all
flows
down
so
I,
don't
know
if
to
tend
is
here,
but
whatever
yeah.
A
All
right
this
is
this
is
good
news
for
actually
these
two
sub
projects
that
we
have
with
windows.
D
D
There
was,
there
was
an
issue
that
came
up
about
using
the
different
load,
balancing
algorithms
that
are
inside
the
hns
load,
balancer.
D
A
D
D
So
it's
a
thing
that
makes
it
so
that
it
can
do
this
firewall
type
Behavior,
where,
if
you
come
in
from
a
certain
IP
it'll
block
you
it's
a
really
weird
Behavior,
though,
because
the
coupe
proxy
isn't
really
supposed
to
be
a
firewall.
It's
just
some
weird
Legacy
feature
that
got
built
in
there
and
that
hasn't
officially
gone
away
or
anything
Windows
is
never
implemented,
and
it
somebody
filed
an
issue
about
it.
D
So
I
confirmed
it's
not
implemented
at
all,
but
then
I'm
reading
leading
the
issue.
So
then,
after
that
David
shot
chimed
in
and
said
that
hns
actually
does
have
this
concept
of
ACLS
that
allow
you
to
block
traffic,
which
I
didn't
know,
but
in
the
hns
API
you
can
do
firewalling.
B
D
D
So
I
don't
think
anybody
has
ever
implemented
this
feature
for
Coupe
proxy,
because
I
don't
know
of
any
other
windows.
Proxy
implementations,
I,
know
of
Andrea
proxy
I
know
of
the
kernel
space
proxy.
That's
in
tree
I
know
this
Windows
service
proxy
that
we're
hacking
up
and
those
are
kind
of
the
only
three
I
know
of
so
I.
Don't
think
anybody's
ever
implemented
this.
So
my
question
to
the
Sig
is:
do
we
want
to
implement
this?
D
And
if
so,
does
anybody
know
anything
about?
No,
no
enough
about
hns
load
balancing
to
suggested
the
right
way
to
implement
this?
B
I
mean
David
and
servant
are
probably
the
best
consult
on
that.
Okay.
C
D
D
It's
mostly
a
matter
of
just
reading
this
reading,
the
incoming
IPS
and
parsing
out
the
cider
that
you're
allowing
and
then
just
comparing
them.
So
it's
not
like
it's
a
super
super
complicated
I
think
there's
a
decent
amount
of
code.
You
can
copy
paste
from
the
existing
Coupe
proxy
to
get
it
working
so
we'll
keep.