►
From YouTube: Kubernetes SIG Windows 20230627
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
Hi
folks
welcome
to
the
June
27th
Sig
Windows
community
meeting.
This
is
a
recorded
meeting
and
will
be
uploaded
to
YouTube.
So
please
add
here
to
the
cncf
code
of
conduct
which,
if
folks
are
not
aware,
translates
to
be
nice
and
kind
to
everyone
all
right.
Let's
kick
things
off
hope
everyone
is
able
to
see
the
agenda
announcements
just
a
reminder
again.
128
code
priest
is
July
19th.
So
it's
coming
up
pretty
quickly.
A
A
Anyone
today
is
a
very
likely
admit,
meeting
everyone
that
I
see
on
the
call
has
introduced
themselves
in
the
past,
so
I
will
move
on
agenda
so,
given
that
we're
so
close
to
the
core
trees,
I
thought
it'll
be
useful
for
us
to
just
go
through
and
ensure
that
the
PRS
that
are
near
and
dear
to
us
is
on
track
to
be
marched
into
128..
A
So
the
first
one
is
I.
Think
fabian's
ADD
end-to-end
test
for
the
Nola
query
feature
Fabian
is
on
the
call
so
what's
needed
for
this
baby
and
I
I
did
a
review,
I
I
remember
doing
a
review
a
while
ago.
Is
it
waiting
on
somebody
else's
reviews
or.
C
B
A
D
Yeah
we
have,
we
can
pass
in
the
feature
gate
to
the
like
as
an
environment
variable
and
it
will
set
the
cluster
up
with
that.
A
Okay,
so
maybe
Fabian
you
could
maybe
follow
up.
Would
that
be
a
follow-up
PR
James.
D
Yeah,
if
we
I
would
say,
let's
merge
them
and
get
him
running,
because
I
think,
if
that
way,
we
can
iterate
on
it,
whereas
what
we're
kind
of
stuck
with
a
couple
of
other
PRS,
because
we
try
to
get
like
a
PR
open
to
validate
the
pr
and
yeah
so
I
think,
and
it's
mostly
that
the
test
trip
is
kind
of
stuck.
So,
okay,
that's,
okay!
We,
if
we,
if
that's
okay
with
folks.
A
Yeah
I'm
fine
getting
this
merged,
and
then
we
can
figure
out
how
to
get
them
running
in
a
follow-up
VR.
D
A
Yeah
two
things:
there's
one
is
the
feature
gate
there's
the
other
option
which
defaults
to
true
that
enable
system
log
query
is
dependent
on,
so
you
basically
need
the
feature
gate
and
an
extra
cubelet
config
option.
D
Yeah,
we'll
have
to
tweak
the
scripts
to
be
able
to
pass
extra
cubelet
parameters,
but
I
think.
D
Because
we're
just
using
cluster
API,
so
we
can
just
use
the
cubelet
extra
arcs
I.
Think:
okay,
I.
A
E
We
can
run
for
the
most
part.
It
seems
that
they're
not
performance,
so
they
won't
typically
be
run
by
other
CIS
and
also,
it
also
depends
if
you
have
some
sort
of
header
for
this
and
in
the
glyco
describe
now.
If
you
have
something
like
feature
columns,
something
something,
those
are
typically
excluded
by
most
of
the
CI
jobs
as
well.
A
A
A
Yeah
so
I
think
we
should
be.
It
should
be
safe.
Then,
based
on
what
Claude
was
saying
that
we
could
get
this
merged,
so
I
I
give
my
lgtm.
It
doesn't
mean
much
because
I
think
the
node
folks
need
to
get
it
merged.
A
I
can
also
you
know
nudge
Bernal
a
little
bit
to
see
if
he
can
get
this
merged.
A
Doesn't
break
yeah?
That's
fine,
all
right,
all
right!
Fabian!
Let
me
know
if
you
need
any
help
with
this
get
getting
merged,
but
we
should
try
and
Target
to
get
this
in
and
then
follow
up
with
the
stuff
that
James
was
saying
all
right.
Let's
move
on
to
crr
CRI
only
stats.
C
A
D
D
This
is
actually
it's
a
bugging
exchanger
D
for
both
Linux
and
windows.
When
the
containers
are
starting
up,
it
was
in
a
if
the
container
isn't
ready.
We
throw
a
hard
error
and
keep
it
was
taking
that
hard
error
and
then
crashing
and
saying,
like
this
doesn't
work,
and
so
this
should
soften
that
so
that
we
can
list
all
the
stats
in
if
container
is
not
quite
ready,
it'll
just
skip
it
move
on
and
the
next
time
we
query
it
it'll
be
there
foreign.
A
D
Oh
so
I
the
we're
running
the
we're
running
the
tests
against
the
main
branch
of
container
D.
Okay.
So
as
soon
as
this
merges,
we
should
be
able
to
rerun
that
test.
Okay,
that
should
we
should
be
able
to
validate
that
the
HCS
shim
one
was
as
a
very
specific
bug
and
I
wasn't
seeing
it
when
I
run
these
two
sets
of
tests,
and
so
we
should
be
okay
to
get
that
coming
in
and
when
the
next
HCS
shim
bumps
bumps.
A
Up
so,
okay
and
there's
nothing
controversial
about
this
PR
right
James.
It
should.
D
No
I
I
I'm,
just
waiting
for
I
got
some
feedback.
They
wanted
me
to
make
it
a
little
bit
more
generic
for
Linux
as
well,
and
so
I
pushed
that
change
up
and
so
I'm
just
waiting
for
them
to
take
a
look
again.
I've
handed
this
off
to
mark
because
I
will
be
out
for
a
little
while
okay,
so
he
does
have
access
to
be
able
to
make
any
changes.
If
anything
comes.
A
Up:
okay,
oh
I,
guess
you
gave
him
access
po
Fork,
I,
guess,
yeah!
Okay,
all
right
all
right
sounds
good.
Looks
like
we're
on
top
of
this
one.
Let's
look
at
vertical
part:
Auto
scaling
for
Windows.
Is
this
Fabian
also
yeah
yeah?
It's
Fabian,
oh
good
baby.
B
I
believe
this
just
needs
the
last
test
to
run
I'm,
not
sure
what
the
errors
are.
I
talked
with
James
I
think
that's
the
only
thing
which
is
skipping
it
from
merging.
B
D
Yeah
so
I'm,
looking
that's
I'm
running
the
test
right
now,
so
I
should
have
an
answer.
Reasons
awesome
all.
D
A
D
Yeah,
so
I
can
speak
to
this
one,
so
we
found
so
this
actually
comes
to
your
if
you're
doing,
HPA
you'll
feel
you
might
run
into
this
eventually.
D
That
limit
is
measured
in
commit
memory,
but
the
cubelet
reports
back
working
set
memory
and
those
two
can
have
a
mismatch,
and
so
in
some
very
small
edge
cases,
you
can
end
up
getting
in
out
of
memory
before
you
scale
so
like
you
should
scale
before
you
hit
out
of
memory,
but
you
might
hit
this,
and
so
this
adds
a
different
metric
specifically
for
Windows
that
will
use
the
commit
memory,
value
and
measure
and,
and
that
just
exposes
it.
So
this
is
the
first
step.
D
The
next
step
is
to
kind
of
switch
HPA
and
everybody
else
to
use
that
level,
and
so
we
are
I
think
we're
good
to
go
here,
we're
just.
We
were
having
a
little
discussion
on
whether
or
not
we
should
change
a
couple
comments,
so
I
think
we're
pretty
close
to
merging
on
this.
D
D
A
Okay,
yeah
I,
don't
see
this
people
agreeing
to
get
this
backboarded
unless
it's
there's
a
big
bug
somewhere
that
it's
fixing
all
right
looks
like
at
least
the
PRS
that
we
are
trying
that
are
mentioned
here
seems
to
be
in
good.
All
seem
to
be
in
good
shape.
To
folks.
Do
folks
have
anything
any
other
PRS
that
they
would
really
like
to
see.
I
know
there
were
a
bunch
of
unit
tests
that
clado
was
working
on
I
didn't
put
them
out
here.
Clutter.
A
E
We
just
have
to
refresh
them
for
a
bit,
but
they
are
just
a
couple
of
small
requests
which
haven't
merged
yet
so
yeah
should
be
fine
too
much
yeah.
E
Approval
Miner
comments
to
be
addressed,
maybe
okay
or
if
we
actually
won
I
mean
a
couple
of
other.
You
need
to
switch
sorry.
There
are
a
couple
of
other
units
which
we
could
fix,
which
flake
so
that's
something
that
we
can
also
look
into,
but
for
the
most
part,
not
a
great
deal
of
work.
There.
A
If
folks
are
running
at
issues,
we
should
start
threads
on
slack
and
you
know
we
should
basically
ensure
that
we
get
these
PRS
in
for
128,
especially
given
at
the
state
where
they
are
not.
Nothing
looks
controversial.
Just
you
know
shaking
some
trees,
making
sure
people
put
the
right
runes
on
the
PRS.
So
let's
let's
try
and
get
them.
It's
done.
That's
what
I
would
suggest
all
right
looks
like
we
are.
We
have
come
to
the
end
of
our
agenda.
Are
there
any
other
topics
that
folks
would
like
to
attend?